JP2006217569A - Image code string conversion apparatus, image code string conversion method, and image code string conversion program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress a reduction in encoding efficiency at the time of re-encoding a moving image code string obtained by encoding a signal of an interlaced scanning method and reduce a calculation amount required for conversion.
A decoding unit generates a decoded image and an encoding parameter from a first moving image code string. In the encoding parameter conversion unit, first, an encoding structure for a set of upper and lower macroblocks is set based on the encoding structure of the corresponding macroblock, and the remaining parameters such as motion vectors are converted based on the encoding structure. The encoding parameter generation unit generates an encoding parameter such as an intra-frame encoding parameter, the encoding parameter selection unit selects an optimum parameter, and the encoding unit performs re-encoding processing.
[Selection] Figure 1

Description

  The present invention relates to an image code string conversion apparatus, an image code string conversion method, and an image code string conversion program.

(Conventional technology)
In the conventional moving image code string conversion device, as disclosed in Patent Document 1, the input first moving image code string is converted into a second moving image code string having a different bit rate, encoding method, or the like. In this case, the encoding parameters included in the first video code sequence are extracted, and the encoding parameters suitable for re-encoding to the second video code sequence are adaptively selected and used. Some have reduced the amount of calculation required for encoding.

(problem)
MPEG-2 (Moving Picture Experts Group Phase 2) video system, MPEG-4 visual system, ITU-T (International Telecommunication Union-Telecommunication Standardization Sector) There are H.264 systems that are international standards. In these encoding methods, one picture is divided into processing units called macroblocks having a defined size, and encoding parameters such as motion vectors and encoding modes are set for each macroblock.

  When an interlaced scanning input signal is encoded by the above encoding method, the handling method may differ depending on the encoding method. For example, in the MPEG-2 video system, encoding can be performed by switching the field structure and the frame structure for each picture or for each macroblock.

  On the other hand, in the H.264 scheme, when switching the coding structure in units of pictures, it is the same as MPEG-2, but in the case of switching the coding structure in units of macroblocks, MB-AFF (Macroblock-Adaptive Frame-Field Coding ) Is used.

  In this MB-AFF encoding method, two macroblocks arranged vertically in the vertical direction are managed together in a processing unit called a macroblock pair, and the field structure and frame structure are switched in units of the macroblock pair. Is called. For this reason, the upper and lower macroblocks included in the macroblock pair must have the same coding structure, and the top field and bottom field coding parameters must be managed separately in the upper and lower macroblocks. Several constraints occur.

For this reason, when converting an input signal that is an interlaced scanning method from a moving image code sequence compressed and encoded in MPEG-2 to a moving image code sequence in the H.264 encoding format, In order to individually set the macroblock coding parameters, it was not possible to efficiently set the macroblock pair coding parameters while satisfying the above-mentioned constraint conditions in the macroblock pair.
JP 2003-9158 (page 7, FIG. 1)

  As described above, the conventional technique has a problem in that the encoding parameter for the macroblock pair in the second video code string cannot be efficiently converted from the macroblock encoding parameter in the first video code string. It was.

  The present invention has been made to solve the above-described problems of the prior art, and corresponds to the macroblock pair when determining the encoding parameter of each macroblock pair in the second moving image code sequence. , By detecting a plurality of macroblocks in the first video code sequence and setting the encoding parameters of the macroblock pair at a time using the encoding parameters of the corresponding macroblock, information on the first video code sequence Is applied to the encoding of the second moving image code string more efficiently, and a moving image code string conversion device, a moving image code string conversion method, and a moving image code string conversion program capable of realizing high encoding efficiency are provided. For the purpose.

In order to achieve the above object, an image code string conversion device according to one aspect of the present invention provides:
An apparatus for converting a first code string obtained by compressing and encoding a video signal of an interlaced order scanning method in units of macroblocks into a second code string based on an encoding method in which compression encoding is performed in units of macroblocks,
A decoding unit that decodes the input first code string to obtain a decoded image and a first encoding parameter;
For each set of macroblocks related to the second code sequence adjacent in the vertical direction on the decoded image, the first of each of the plurality of corresponding macroblocks related to the first code sequence corresponding to the set on the image. An encoding parameter conversion unit that converts one encoding parameter to obtain a conversion encoding parameter;
An encoding parameter selection unit that selects the transform encoding parameter obtained for the set as a second encoding parameter of each macroblock of the set;
An encoding unit that compresses and encodes the decoded image using the second encoding parameter selected by the encoding parameter selection unit and generates the second code string;
Is provided.

  According to the present invention, the encoding parameters for the macroblock pair in the second video code sequence are collectively set based on the encoding parameters of the corresponding macroblock in the first video code sequence, so that the macro block pair While satisfying the constraint condition of the encoding parameter, the encoding parameter can be efficiently reused, and the encoding efficiency can be improved and the deterioration of the image quality can be suppressed.

  Hereinafter, embodiments of the present invention will be described.

(First embodiment)
FIG. 1 is a block diagram showing a moving image code string conversion apparatus according to the first embodiment of the present invention.

The moving image code string conversion device according to the first embodiment is
A decoding unit 110 that decodes the first video code sequence 100 as an input, generates a decoded image 101, and extracts a coding parameter 102;
An encoding parameter conversion unit 111 that analyzes the encoding parameter 102 and converts the encoding parameter in accordance with the encoding format of the second moving image code sequence;
An encoding parameter generation unit 113 that generates an encoding parameter candidate 104 using the decoded image 101 and the already encoded second video code sequence 103;
An encoding parameter selection unit 112 that finally selects a parameter 106 used for encoding from the encoding parameter 105 converted by the encoding parameter conversion unit 111 and the encoding parameter 104 generated by the encoding parameter generation unit;
An encoding unit 114 that generates a second moving image code sequence using the encoding parameter 106 selected by the encoding parameter selection unit 112.
And.

  In this embodiment, in particular, the first moving image code string is obtained by encoding a moving image signal of interlace scanning using the encoding method MPEG-2. A case will be described in which MPEG-4 AVC is used as the second moving image encoded signal.

  Next, the operation of the moving picture code string conversion apparatus according to the first embodiment of the present invention will be described with reference to FIGS. FIG. 2 is a flowchart showing the operation of the video code string converter according to the first embodiment of the present invention.

  First video code sequence 100 is input to decoding section 110 (step S100).

  In the decoding unit 110, the first moving image encoded sequence 100 is decoded to generate an encoding parameter 102 and a decoded image 101 (step S101).

  The encoding parameter conversion unit 111 converts the encoding parameter 102 of the first moving image code sequence 100 into the second encoding parameter 105 (step S102).

  The encoding parameter generation unit 113 generates the encoding parameter 104 using the decoded image 101 of the first moving image code sequence and the already encoded second moving image code sequence 103 (step S103).

  The code parameter selection unit 112 evaluates the parameter 105 converted by the encoding parameter conversion unit and the parameter 104 generated by the encoding parameter generation unit, and finally selects the parameter 106 used for encoding (step S104).

  Using the parameter 106 selected by the encoding parameter selection unit, the encoding unit 114 encodes the decoded image of the first video code sequence into the second video code sequence 103 (step S105), and the second video code The column 103 is output (step S106).

  Note that the encoding parameter 107 obtained by converting the encoding parameter extracted from the first moving image encoded sequence 100 by the encoding parameter conversion unit 111 is input to the encoding parameter generation unit 113, and the encoding parameter is set. It may be configured to be used as reference data for generation.

  The details of each processing unit in FIG. 1 will be described below.

(Decryption unit)
The decoding unit 110 decodes the input video code sequence 100 and extracts the code parameters 102 included in the decoded image 101 and the first video code sequence.

  The encoding parameter 102 extracted here includes the following information when the first moving image code sequence 100 is MPEG-2.

Resolution information indicating the size of the decoded image included in the code string,
Progressive information indicating whether the entire code string is a sequential scanning method,
Picture type of picture (intra-frame coding (I picture), inter-frame coding (P picture, B picture)),
Picture coding structure (frame, field),
Macroblock coding mode (intraframe coding, interframe coding),
A motion vector indicating a reference destination in macroblock motion prediction,
Block shape during macroblock motion compensation,
DCT type present when the macroblock is intra-frame coded,
The prediction type that exists when the macroblock is interframe coded,
A field select that indicates which field the macroblock motion vector references,
The decoding unit 110 of this embodiment extracts these pieces of information as the encoding parameters 102.

(Encoding parameter converter)
A processing flow of the encoding parameter conversion unit 111 is shown in FIG.

  A macro block corresponding to the macro block pair is obtained (step S110).

  Based on the corresponding macroblock, the coding structure of the macroblock and the macroblock pair, that is, coding with a frame structure or coding with a field structure is determined (steps S111 and S112).

  The determined macroblock pair and each coding structure of the corresponding macroblock are compared (step S113), and if there are contradictory parameters or insufficient parameters, they are converted or generated (step S114).

  The encoding parameter of the corresponding macroblock is converted in accordance with the encoding scheme of the second moving image code string (step S115), and is output as an encoding parameter candidate when generating the second moving image code string.

  Hereinafter, each process of FIG. 3 is demonstrated concretely.

(S110: Detection of corresponding macroblock)
The macro block corresponding to the macro block pair is obtained. In the present embodiment, the resolution conversion process is not included, the first moving image code sequence, the second moving image code sequence, and the resolution are the same, and the size of the macroblock also matches in both encoding methods. As shown in FIG. 4, the two macroblocks 201 and 202 that are the conversion source of the first video code sequence correspond to the macroblock pair 200 that is the conversion destination in the second video code sequence.

(S111: Setting of coding structure of corresponding macroblock)
A coding structure is obtained for each corresponding macroblock. In the present invention, the coding structure is defined as either a field structure or a frame structure.

  A specific method for determining the coding structure is shown in FIG. First, the coding mode of the macroblock is analyzed (step S120), and it is determined whether the coding is intraframe coding (Intra) or interframe coding (NonIntra) (step S121).

  In the case of inter-frame coding, the information on the prediction type included in the macroblock is also analyzed (steps S123 and S125). If the prediction type is field prediction, the field structure (step S127), and if it is frame prediction, the frame structure (Step S126).

  If the coding mode of the macroblock is intraframe coding, the DCT type information of the macroblock is analyzed (steps S122 and S124), and if the DCT type is field DCT, the field structure (step S127) If it is DCT, the frame structure is set (step S126).

(S112: Determination of coding structure of macroblock pair)
After the determination of the coding structure of each macroblock is completed, the coding structure of the macroblock pair is determined based on the coding structure of the corresponding macroblock.

  A specific processing flow is shown in FIG.

  The coding structures of the corresponding two macroblocks are compared (steps S130 and S131).

  When the coding structures of the two macroblocks are equal (when step S131 is Yes), the coding structure is directly used as the coding structure of the macroblock pair. That is, if the coding structure of both macroblocks is a field structure, the field structure (step S133) is used, and if the coding structure is a frame structure, the frame structure is used (step S134).

  If one is a frame structure and the other is a field structure (No in step S131), the coding structure of the macroblock pair is a field structure (step S133).

  When the picture type of the conversion source picture is intra-frame coding, the coding mode conversion process is completed with the coding mode of the macroblock pair as intra-frame coding. When the encoding mode of the conversion source picture is inter-frame encoding, encoding parameters used for predictive encoding are converted as follows.

(S114: Conversion of corresponding macroblock parameters in accordance with the coding structure of the macroblock pair)
The encoding parameter of each macroblock corresponding to the macroblock pair is converted in accordance with the encoding scheme of the first moving image code string so as to match the encoding structure of the macroblock pair.

(Conversion from intra-frame coding to inter-frame coding)
A specific processing flow is shown in FIG. First, for each of the corresponding macroblocks, when the encoding mode of the macroblock is intraframe encoding (when step S141 is Yes), the corresponding macroblock is converted into an interframe encoding mode, A prediction type is also generated based on the coding structure of the macroblock (step S142).

  Specifically, when the DCT type of intra-frame coding has a field structure, the prediction type is set to field prediction, and when the DCT type has a frame structure, it is set to frame prediction. For the motion vector, a flag is set so that motion vector generation by motion vector prediction is performed on the encoding unit side. This is because, in the encoding parameter conversion in the present embodiment, when the picture type of the picture to be converted is inter-frame encoding, all the encoding parameters output as the conversion result are for inter-frame encoding. .

(Conversion of motion vector due to coding structure change)
Next, each coding structure of the corresponding macroblock is compared with the coding structure of the conversion target macroblock pair (step S143). When the coding structures are different (when Step S143 is Yes), the coding structure of the corresponding macroblock, the motion vector, and the prediction type are converted in accordance with the structure of the macroblock pair (Step S144).

  Specifically, when the coding structure of the corresponding macroblock 800 is a frame structure and the coding structure of the macroblock pair 801 is a field structure, the prediction type of the corresponding macroblock 800 is set to a field as shown in FIG. The frame prediction motion vector is converted into a field prediction motion vector. At this time, two field predictions MV2 ′ (topMV2 ′ and bottomMV2 ′) having the same value are generated by halving the vertical component of the frame prediction motion vector MV2.

  When the coding structure of the corresponding macroblock is a field structure and the coding structure of the macroblock pair is a frame structure, a frame prediction motion block is generated from the field prediction motion vector included in the corresponding macroblock. The above conversion process is repeated until the coding structure of the corresponding macroblock matches the coding structure of the macroblock pair (step S145).

(Encoding format conversion)
When the correspondence between the corresponding macroblock and the coding structure of the macroblock is obtained, the motion vector and the motion compensation block shape are finally converted.

  A specific conversion method is shown in FIGS.

  FIG. 9 shows the shapes of the motion compensation blocks 900 and 901 when the coding structure of the macroblock pair is a frame structure. When the coding structure of the macroblock pair is a frame structure, as shown in FIG. 9, the motion compensation block shape is 16x16 in both upper and lower directions, and the motion vector is the same as the frame prediction motion vector in consideration of pixel accuracy. Convert to the motion vector you point to.

  FIG. 10 shows the shapes of motion compensation blocks 1000, 1001, 1002, and 1003 and the assignment of motion vectors when the coding structure of the macroblock pair is a field structure.

  On the other hand, when the coding structure of the macroblock pair is a field structure, the motion compensation block shape is 16 × 8 as shown in FIG. Also, for motion vectors, two motion vectors topMV1 and topMV2 for the top field are assigned to the macroblock above the macroblock pair, and two motion vectors bottomMV1 and bottomMV2 for the bottom field are assigned to the macroblock below the macroblock pair. .

  In addition, about a motion block shape, you may set the parameter regarding the block shape other than the block shape set as mentioned above.

  For example, in the case of a field structure, a 16x16 block shape may be set from two 16x8 block shapes. In this case, a motion vector is selected from among two 16 × 8 block shapes using an evaluation value such as the magnitude of a residual signal.

  In addition, two 16x8 block shapes may be generated from one 16x16 block shape. In this case, the motion vector may be assigned to the two block shapes for the 16x16 block shape, or set to use motion vector prediction, and the motion vector is set in the subsequent encoding parameter generation unit. It may be calculated.

(Encoding parameter generator)
The encoding parameter conversion unit 111 generates the encoding parameter 105 for the second video code sequence from the encoding parameter 102 of the first video code sequence 100, whereas the encoding parameter generation unit 113 Using the decoded image 101 of the first video code sequence and the already encoded second video code sequence 103, an encoding parameter for the second video code sequence is generated.

  For example, when the picture type of the first moving image code sequence is inter-frame coding, the coding parameter unit mainly generates coding parameters for inter-frame coding. Intraframe encoding parameters, skip, and direct prediction data that are not generated by the parameter conversion unit are generated.

  As for the interframe coding parameters, the coding parameter generation unit performs motion vector prediction and generates a motion vector for a portion where a flag is set so that the prediction parameter is used by the coding parameter conversion unit.

(Encoding parameter selector)
FIG. 11 shows a processing flow of the encoding parameter selection unit.

  Conversion parameter candidates 105 and 104 used for conversion are acquired from the encoding parameter conversion unit 111 and the encoding parameter generation unit 113. Specifically, the inter-frame prediction encoding parameter 105 generated by the encoding parameter conversion unit 111, the intra-frame prediction encoding parameter generated by the encoding parameter generation unit 113, and other encoding parameters. Get 104.

  The encoding efficiency is evaluated for each encoding parameter candidate. Specifically, for motion vector evaluation, an absolute sum of differences (SAD), rate distortion characteristics (R-D: Rate-Distortion), or the like can be used.

  Finally, an encoding parameter with the highest encoding efficiency is selected as a parameter used for actual encoding.

(Encoding part)
The encoding unit encodes the decoded image using the parameter selected by the encoding parameter selection unit, and generates and outputs a second moving image code string.

  As described above, according to the moving picture code string conversion apparatus according to the first embodiment, by setting the encoding parameters for the macroblock pair collectively based on the encoding parameters of the corresponding macroblock, The encoding parameters can be reused efficiently while satisfying the encoding parameter constraint conditions in the pair, and the encoding efficiency can be improved and the deterioration of the image quality can be suppressed.

(Second Embodiment)
FIG. 12 is a block diagram showing a video code string conversion apparatus according to the second embodiment of the present invention.

  Compared with FIG. 1 in the first embodiment, a difference is that a resolution conversion unit 120 is placed after the decoding unit 110 and the decoded image 101 is re-encoded after resolution conversion.

  In the present embodiment, the resolution conversion unit converts to a resolution lower than the resolution of the input signal.

  In the encoding parameter conversion unit 111 in the second embodiment, the number of corresponding macroblocks corresponding to one macroblock pair is greater than two. An example specifically showing the relationship with the corresponding macroblock is shown in FIG. In the example of FIG. 13, corresponding macroblocks 301 and 302 occupying 24 macroblocks of the decoded image 101 correspond to the macroblock pair 300.

  Since the field structure is more actively used as the coding structure of the macroblock pair, if the corresponding macroblock includes at least one field structure (MNUM1> 0), the coding structure of the macroblock pair is changed. A field structure may be used.

(Determine coding structure of macroblock pair)
Unlike the processing of FIG. 6 in the first embodiment, is determined based on the number of coding structures of corresponding macroblocks. A specific processing flow is shown in FIG.

  First, as shown in FIG. 13, a plurality of macroblocks (a plurality of correspondences included in the areas indicated by 301 and 302) corresponding to each macroblock pair 300 of the second video codestream are included. Macroblock) is extracted, and the coding structure of each macroblock is grasped (step S170).

  Next, the field structure macroblock number MNUM1 and the frame structure macroblock number MNUM2 are calculated for a plurality of corresponding macroblock coding structures (step S171).

  Next, the number of macroblocks MNUM1 in the field structure is compared with the number of macroblocks MNUM2 in the frame structure (step S172). As a result of this comparison, if the number of macroblocks MNUM2 in the frame structure is greater than the number of macroblocks MNUM1 in the field structure, a second video code string is generated by re-encoding that uses the frame structure (step S173), and the frame structure When the number of macroblocks MNUM2 is smaller than the number of macroblocks MNUM1 of the field structure, a second video code string is generated by re-encoding using the field structure (step S174).

  MNUM1 and MNUM2 are compared, and if the number is the same, a second video code string is generated by re-encoding using the field structure (step S174).

  Also, when the number of MNUM1 and MNUM2 is the same, an evaluation value based on the encoding parameter may be set for each macroblock, and the encoding structure may be determined using the sum of the evaluation values of each encoding structure . The evaluation value includes the code amount of the macroblock and the product of the code amount and the quantization value.

  In the above, the method of determining the coding structure when generating the second video code sequence by simply comparing the number of macroblocks NUM1 of the field structure and the number of macroblocks NUM2 of the frame structure has been described.

  However, for example, the corresponding area of the macroblock having each coding structure corresponding to the macroblock pair of the second moving image code string is obtained, and an evaluation value is set according to the ratio of the corresponding area, and the code of the macroblock pair The coding structure may be determined by determining the coding structure. Hereinafter, a case where such a determination method is applied will be described.

  Here, the case of converting the resolution in the horizontal direction will be described.

  First, it is assumed that four macroblocks correspond to one macroblock pair at a ratio of a: b as shown in FIG.

At this time, the evaluation value V _field related to the field structure and the evaluation value _Vframe related to the frame structure are _obtained by using the following Expression 1 and Expression 2. Here, the ratio of the values of V _field and V _frame represents the ratio of the area corresponding to the macro block having the frame structure to the macro block pair and the area corresponding to the macro block having the field structure.

However, MB ₀ , MB ₁ , MB ₂ , and MB ₃ in Formula 1 and Formula 2 represent the areas of four macroblocks shown in FIG. In addition, T _field and T _frame in Expression 1 and Expression 2 are functions that define whether the coding structure of the target macroblock is a field structure or a frame structure, respectively. Defined by

  FIG. 19 shows a specific processing flow based on this method.

  First, a plurality of macroblocks of the first video code sequence corresponding to each macroblock pair of the second video code sequence are extracted, and the coding structure of each macroblock is grasped (step S180).

Next, an evaluation value V _field related to the field structure and an evaluation value _Vframe related to the frame structure are calculated using Expression 1 and Expression 2 (Step S181).

Next, the values of V _field and V _frame calculated by Equation 1 and Equation 2 are compared (step S182). If V _field <V _{frame as} a result of this comparison, a second video code sequence is generated by re-encoding using the frame structure (step S183). If V _field ≧ V _frame , the second moving image code string is generated by re-encoding using the field structure (step S184).

At this time, the value of the V _field may be weighted so that the field structure is preferentially selected. A specific processing flow based on this method is shown in FIG.

  First, a plurality of macroblocks of the first video code sequence corresponding to each macroblock pair of the second video code sequence are extracted, and the coding structure of each macroblock is grasped (step S190).

Next, an evaluation value V _field related to the field structure and an evaluation value _Vframe related to the frame structure are calculated using Expression 1 and Expression 2 (Step S191).

Next, the value of the weighting coefficient w is set (step S192), and the value obtained by weighting the value of the V _field is compared with the value of the V _frame (step S193). As a result of this comparison, if V _frame > w × V _field , the frame structure is selected (step S194), and if V _frame ≦ w × V _field , the field structure is selected (step S195).

(Adaptive adjustment of weighting factor)
The weight coefficient w is set for each moving image code string to be converted. The value of the weighting factor w may be uniquely determined for each moving image code string, or may be determined adaptively within the moving image code string.

  As an adaptive setting method of the weighting coefficient w, for example, a method of determining based on a conversion ratio of resolution, a method of determining based on a ratio between the bit rate of the first video code sequence and the bit rate of the second video code sequence The method to do can be applied. Further, the value of the weighting factor w may be uniquely determined for each moving image code string in accordance with the magnification at the time of resolution conversion, or information on the corresponding macroblock (the size of the motion vector, the variance, the code amount, etc.) Depending on the information of the picture (picture type, code amount, etc.), it may be changed adaptively.

(Weight factor setting method: always use a constant value)
Hereinafter, a specific example of the method for setting the weighting coefficient w will be described.

When setting the weighting factor w so that the field structure is always obtained when the area ratio of the macroblocks of the field structure corresponding to the macroblock pair exceeds a certain ratio α (0 ≦ α ≦ 1) The coefficient w can be calculated by the following formula 5.

  By using the weighting coefficient w defined in the above equation 5, when a macroblock having a field structure is always included in a certain ratio α or more, the coding structure of the macroblock pair can be determined as the field structure.

(Weight coefficient setting method: Use distance of motion vector)
Next, as another setting method of the weight coefficient w, a setting method based on the difference between the vector corresponding to the top field and the vector corresponding to the bottom field in the motion vector of the field prediction will be described below.

  Consider a case where a macroblock having a field structure exists in a macroblock corresponding to a macroblock pair.

  In the case of a macroblock that is field prediction, there are a vector corresponding to the top field and a vector corresponding to the bottom field. At this time, if the distance between the two motion vectors included in the macroblock is large, it is considered that there is a high possibility that two fields having different phases exist in the corresponding macroblock. The Euclidean distance or the like is used as the distance between vectors at this time.

  Therefore, by changing the weighting factor w adaptively according to the distance between the two field prediction motion vectors included in this one corresponding macroblock, the coding structure in accordance with the state of the source MPEG-2 is converted. Decisions can be made.

  At this time, the distance between the field prediction motion vectors may be calculated in the same manner for the macroblock having each field structure of the entire picture, and the value of the weight coefficient w may be changed in units of pictures based on the statistics.

(How to set weighting factor: use bit rate)
When MPEG-2 frame prediction and field prediction are compared, frame prediction with a 16x16 block shape can achieve substantially the same motion compensation with two 16x8 field predictions. The amount of code increases as the motion vector increases, but if the bit rate after conversion is large enough to ignore the increase in the amount of motion vector code, the field structure should be prioritized and the conversion source MPEG-2 In many cases, the motion vector can be used effectively.

  On the other hand, when the bit rate after conversion is small and the ratio of the code amount of the macroblock header portion to be encoded is large, the code amount of the motion vector may greatly affect the overall image quality. For this reason, the image quality is improved by changing the value of the weighting factor w for the field structure in accordance with the ratio of the bit rate of the converted second video code sequence to the bit rate of the first video code sequence of the conversion source. Is possible.

  The value of the weighting factor w corresponding to the bit rate ratio may be uniquely determined for the stream. In addition, a variable bit rate stream may be dynamically changed with finer accuracy based on a code amount for each GOP (Group of Picture) or each picture.

(Weight factor setting method: Use of activity)
Next, a case where the value of the weighting coefficient w is adjusted based on the activity of the moving image code string will be described.

  Here, the activity is a numerical value representing the information amount of the moving image to be encoded. In the moving image conversion process, the activity is calculated as follows.

  The activity in units of macroblocks is obtained by multiplying the code amount for each macroblock by the value of the quantization step width. In the case of a picture unit, it is obtained by multiplying the average quantization step width of a picture and the picture code amount.

Using the activity calculated as described above, the value of the weighting factor w is actually dynamically changed. For example, the activity values of the field structure and frame structure macroblocks are _added to the area ratio in the same manner as the evaluation values V _field and V _frame, and the weight coefficient w is adjusted using the ratio. It is possible. Further, when the values of V _field and V _frame are close, the final coding structure of the macroblock pair may be determined based on the magnitude of the activity of each coding structure. In addition, the value of the weighting factor w may be adjusted adaptively for each picture based on the activity for each picture.

(Weight coefficient setting method: Allow multiple method combinations)
About the adjustment method of the weighting factor mentioned above, you may use individually, respectively, and you may use combining several adjustment methods.

(Weighting factor is used as an offset value)
Further, the weight coefficient w and V _field may be added together as an offset value instead of linearly combining. Specifically, if V _frame > w + V _field , the frame structure is used, and if V _frame ≦ w + V _field , the field structure is used. Also in this case, the value of the weighting factor w can be changed using the same method as described above. FIG. 21 shows the flow of processing.

  First, a plurality of macroblocks of the first video code sequence corresponding to each macroblock pair of the second video code sequence are extracted, and the coding structure of each macroblock is grasped (step S200).

Next, the evaluation value V _field related to the field structure and the evaluation value _Vframe related to the frame structure are calculated using Expression 1 and Expression 2 (Step S201).

Next, the value of the weighting factor w is set (step S202), and the value obtained by weighting the value of the V _field is compared with the value of the V _frame (step S203). As a result of this comparison, if V _frame > w + V _field , the frame structure is selected (step S204), and if V _frame ≦ w + V _field , the field structure is selected (step S205). The weighting factor w can be adjusted uniquely or adaptively using the same method as described above.

(Vertical direction, support for 4 or more macroblocks)
In the above description, the case where four macroblocks correspond to one macroblock pair is described only when the resolution conversion is performed in the horizontal direction. However, when the resolution conversion in the vertical direction is performed, horizontal and vertical When performing resolution conversion in both directions, or when more than four macroblocks correspond to one macroblock pair, the corresponding ratio is obtained for each macroblock in the same manner as described above, and an equation is determined according to the corresponding ratio. By extending 1 and Equation 2, it is possible to determine the coding structure for generating the second moving image code string in the same manner as described above.

(Motion vector conversion)
At the time of motion vector conversion, there are a plurality of motion vector candidates in the conversion accompanied by resolution conversion. Therefore, which motion vector is suitable is compared using an evaluation value, and an optimal motion vector is selected.

  As the evaluation value, for example, a prediction residual code amount of a macroblock, a quantized value, a motion vector size, a multiplication value of the prediction residual and the code amount, and the like can be used.

  Also, the motion vector is scaled according to the resolution conversion ratio. Note that, at the time of scaling the motion vector, the fraction processing is performed with the accuracy supported by the encoding scheme of the second moving image code string. Specifically, motion vector accuracy is 1/2 pixel accuracy with MPEG-2, but 1/4 pixel accuracy with H.264, so values are obtained with 1/4 pixel accuracy during scaling. .

(Conversion of motion compensation block shape)
Unlike the first embodiment, the motion compensation block shape may not be converted in one-to-one correspondence. In addition, since the size of the macro block of the first video code sequence corresponding to the macro block pair becomes relatively small with the resolution conversion, even if any one block shape of the corresponding macro block is used, the block There may occur a situation where the shape becomes too small and the coding efficiency decreases.

  In such a case, the motion compensation block shape is determined in advance based on the coding structure of the macroblock pair. Specifically, a 16 × 16 block shape is used for the frame structure, and a 16 × 8 block shape is used for the field structure.

  Note that the block shape to be generated need not be limited to one as in the case of the first embodiment. For example, a 16 × 8 block shape may be generated with a frame structure, or a 16 × 16 block shape may be generated with a field structure.

  In the final conversion, as in the first embodiment, for each corresponding macroblock, parameters are generated as shown in FIG. 2 according to the macroblock pair structure, and the code for the second video code string is generated. Convert to conversion parameter.

  The processes of the decoding unit, the encoding parameter selection unit, and the encoding unit are performed in the same manner as in the first embodiment.

  As described above, according to the moving image code string conversion device according to the second embodiment, the encoding parameter for the macroblock pair is based on the encoding parameters of the corresponding macroblocks even in the conversion with resolution conversion. By setting all together, it is possible to efficiently reuse the coding parameters while satisfying the restriction condition of the coding parameters in the macroblock pair, and further reduction of the code amount can be expected.

(Third embodiment)
The configuration and operation of the third embodiment of the present invention will be described with reference to FIGS. FIG. 15 is a block diagram of a video code string converter according to the third embodiment of the present invention.

  In the third embodiment of the present invention, a first video code sequence 500 encoded by the MPEG2 system is input, and a second video code sequence 503 re-encoded by the H.264 system is output. It is.

  The first moving image code string 500 is decoded by the MPEG2 decoder 510 to output a decoded image 501, a picture type (I picture, P picture, or B picture), and a coding sequence of a progressive picture. Encoding in units of encoded pictures, such as a flag indicating whether it is encoded only (hereinafter referred to as progressive sequence) or encoded data including an interlaced image, picture structure (progressive frame picture, interlaced frame picture, field picture) The parameter (the upper stage of 502) is output, and the picture type setting unit 522 determines the picture level coding structure at the time of H.264 re-encoding.

  FIG. 16 shows an operation example in the picture type setting unit 522 according to the third embodiment of the present invention. When MPEG2 encoded data is encoded as a progressive sequence, each corresponding frame is re-encoded as a progressive frame picture in H.264.

  When the MPEG2 encoded data is a non-progressive sequence, the MPEG2 frame picture is replayed as an H.264 MB-AFF frame picture, and the MPEG2 field picture is replayed as an H.264 field picture. Encode. In addition, a picture whose type of coded picture in MPEG2 is an I picture is re-encoded as a reference picture composed of only I slices in H.264, and an MPEG2 B picture is composed only of B slices in H.264 Re-encoding as a non-reference picture.

  Furthermore, for MPEG2 P-pictures, dual-prime prediction defined in MPEG2 (a mode in which predictive image blocks are cut out from the reference pictures of the two preceding fields and inter-picture predictive coding is performed using the average value as a predictive image) ) Are re-encoded as a reference picture made up of only H.264 B slices.

  In addition, an MPEG2 P picture that does not include dual-prime prediction is re-encoded as a reference picture composed of only H.264 P slices. By providing the picture level correspondence as described above, it becomes easy to associate MPEG2 and H.264 of the inter-frame prediction structure in macroblock units, and reuse of motion vectors and prediction modes in macroblock units. Thus, it is possible to significantly reduce the calculation cost (calculation amount or hardware cost) at the time of H.264 re-encoding.

  FIG. 17 shows another embodiment of the picture type setting unit 522. The difference from the configuration of FIG. 16 is that an MPEG2 P picture is re-encoded as a reference picture composed of only H.264 P slices regardless of the presence or absence of dual-prime prediction. In an MPEG2 P picture, dual-prime prediction can be used only when a B picture is not included in the playback order between the P picture and its reference picture, and prediction of all macroblocks in the picture If the mode is not examined, the presence or absence of dual-prime prediction cannot be determined.

  Accordingly, in the re-encoding from MPEG2 to H.264, a delay of at least one frame is required, and an increase in buffer memory necessary for the delay and a time delay associated with the re-encoding occur.

  On the other hand, since dual-prime prediction performs average prediction of the past two fields, it is impossible to perform similar inter-picture prediction in the H.264 P slice. Also, if all MPEG2 P pictures are re-encoded as H.264 B slices, the overhead of encoded data indicating the prediction mode in units of prediction blocks of H.264 B slices will increase, and the coding efficiency will be increased. It will cause a decline.

  In order to solve these problems, in macroblocks using MPEG-2 dual-prime prediction, as a single prediction using only the reference field temporally closest to the current picture to be encoded among the reference pictures of the past two fields By re-encoding with H.264, it becomes possible to re-encode all macroblocks of MPEG2 P-pictures with H.264 P-slices by reusing the motion vector information. . This solves the problem of delay at the time of re-encoding and the problem of reusability of the prediction structure together, suppresses the increase in calculation cost at the time of H.264 re-encoding, and further encodes at the time of re-encoding. It is possible to prevent a decrease in efficiency.

  Next, individual operations in units of macroblocks in FIG. 15 will be described.

  Of the encoding parameters for each macroblock in the first moving image code string 500, the quantization scale value and the code amount for each macroblock (the middle stage of 502) are extracted and input to the rate control unit 523, where The generated code amount in units of encoding macroblocks is also fed back to the rate control unit 523 and used for code amount control during H.264 re-encoding.

  Although it is not indispensable to input the encoding information in the first moving image code sequence 500 for the code amount control at the time of re-encoding, by using the information of these MPEG2 encoded data, It is possible to grasp the conversion characteristics in advance and realize efficient rate control with little image quality degradation.

  In addition, among the coding parameters in units of macroblocks in the first moving image code string 500, motion compensation modes (frame prediction, field prediction, dual-prime prediction, etc.) defined by the MPEG2 standard, DCT type (frame DCT) Or field DCT) is extracted and input to the encoding parameter conversion unit 520. If the encoding parameter is directly compatible with H.264, it is reused as it is when re-encoding.

  For example, when MPEG2 is encoded as a progressive sequence, the motion vector of MPEG2 can be used as it is by setting the motion compensation block size of H.264 to 16 × 16 pixels.

  Also, when encoded as a field picture in MPEG2, the 16x16 field prediction mode of MPEG2 is 16x16 prediction of H.264, and the 16x8 field prediction mode of MPEG2 is also between 16x8 size pictures in H.264. As a prediction, the motion vector can be used as it is.

  In the case of MPEG-2 dual-prime prediction, when re-encoding as an H.264 B slice as described above, the motion vector can be reused as it is and re-encoded as an H.264 P slice. In this case, the motion vector can be reused by performing the single prediction only from the reference field close to the encoding target picture.

  Also, if the MPEG-2 encoding parameter is not directly compatible with H.264, such as the MB-AFF pair type, the MPEG2 encoding is performed as described in the first or second embodiment of the present invention. Appropriate conversion is performed on the parameter by the encoding parameter conversion unit 520, and further, the frame / field setting unit 521 sets whether the encoding structure of the macroblock pair unit at the time of H.264 re-encoding is a frame structure or a field structure And reused in H.264 re-encoding.

  More specifically, an MPEG-2 I picture is re-encoded as a picture composed of only H.264 I slices, and when a frame DCT and a field DCT are mixed in an MPEG2 I picture, MB-AFF Re-encode as a frame.

  When MPEG2 macrobook pairs corresponding to H.264 macroblock pairs are encoded with the same DCT type, the macroblock pair type (frame pair or field pair) is determined according to the type. A pair in which frame DCT and field DCT are mixed is a field pair.

  The type of macroblock pair determined above, and the intra-prediction coding unit 525 may select one of a plurality of intra-prediction modes (Intra4x4 prediction (maximum 9 modes), Intra8x8 prediction (maximum 9 modes), Intra16x16 prediction (maximum 4 modes)) Thus, the optimum prediction mode set is determined for each macroblock and re-encoding is performed.

  The optimum prediction mode is determined using an evaluation value such as a sum of absolute differences of prediction error signals, a generated code amount in each mode or an estimated value thereof, and a rate − using an encoding distortion or an estimated value thereof. It can be determined by mode determination using a distortion optimization method.

  By determining the MB-AFF pair type in advance from only MPEG2 encoding information as described above, the process of re-determining the optimal pair type in H.264 re-encoding becomes unnecessary, and re-encoding is not necessary. Calculation costs (calculation amount or hardware cost) can be greatly reduced.

  In addition, for P-pictures or B-pictures in MPEG2, macroblocks for which inter-picture predictive encoding is used are extracted from MPEG2 encoded data and centered on motion vectors that are appropriately converted as necessary. Then, the motion vector re-search unit 524 re-searches the motion vector in the minute region as necessary.

  MPEG2 normally uses a motion vector with 1/2 pixel accuracy, and H.264 normally uses a motion vector with 1/4 pixel accuracy. Therefore, in this embodiment, the re-search of the motion vector is configured to search for at least the vicinity ± 0.25 of the search center.

  In addition, in order to improve the encoding efficiency at the time of re-encoding, a more appropriate motion vector can be obtained by rounding the search center of 1/2 pixel accuracy to an integer pixel position and searching for a range of ± 1.75 in the vicinity thereof. It becomes possible to decide.

  This is because the search center obtained from the motion vector with 1/2 pixel accuracy of MPEG2 is not necessarily optimal because the characteristics of the interpolation filter for MPEG2 decimal pixel position prediction and the H.264 decimal pixel interpolation filter are different. It is. The motion vector search center rounded to the integer pixel position has an error of ± 1 pixel with respect to the optimum value of the integer precision motion vector, and the effect of the interpolation filter varies in the range of ± 0.75 pixel. By searching a range of ± 1.75 pixels from the rounded search center, an appropriate motion vector can be obtained. This makes it possible to perform motion vector detection for H.264 re-encoding that gives high encoding efficiency by re-searching a minute region of ± 1.75 pixels, reducing re-encoding costs and encoding It is possible to improve both efficiency.

  In addition, the motion vector re-search result and the intra prediction encoding unit 525 can select one of a plurality of intra prediction modes (Intra4x4 prediction (maximum 9 modes), Intra8x8 prediction (maximum 9 modes), Intra16x16 prediction (maximum 4 modes)). Therefore, an optimal set of prediction modes is determined for each macroblock, and H.264 inter-picture prediction mode with high coding efficiency that does not require coding of motion vectors, and P_Skip in H.264 for P slices. For modes and B slices, the H.264 direct mode predictive coding is performed by the Skip / Direct coding unit 526, and the coding mode with the highest coding efficiency is selected from these for each macroblock. The determination unit 512 makes a selection.

  Here, for intra-coded macroblocks in P-pictures or B-pictures in MPEG2, using motion vectors that have been re-searched using the prediction vector in motion vector predictive coding specified in H.264 or the prediction vector as a search center Then, predictive encoding is performed by the inter-picture prediction unit 527, and the output is set as a candidate for the inter-picture encoding mode.

  With the above configuration, in a P picture or B picture in MPEG2, regardless of whether the encoding mode of the macroblock is intra encoding or inter picture predictive encoding, H having a high encoding efficiency having a large number of prediction modes. Select the optimal mode from .264 intra coding, P_Skip or direct prediction, which is an inter-picture prediction with high coding efficiency, and a motion vector search result in an appropriate range using MPEG2 motion vectors. As a result, it is possible to greatly improve the encoding efficiency at the time of re-encoding while suppressing the calculation cost.

  Here, in order to reduce the processing cost of the coding mode determination, when selecting an optimal mode from a plurality of H.264 intra prediction modes or re-searching motion vectors, the amount of computation such as the sum of absolute values of prediction residual signals Mode evaluation is performed with a small evaluation value, and at the macroblock level mode determination at the final stage, the rate-distortion optimization cost based on the generated code amount and coding distortion, or the estimated generated code amount or estimated coding distortion By performing high-accuracy mode determination using the coding cost based on the evaluation value, it is possible to select a suitable coding mode and improve the coding efficiency while suppressing an increase in the amount of calculation.

  Finally, an H.264 code example is generated by an H.264 CABAC (Context Adaptive Binary Arithmetic Coding) coding unit 528 in the selected coding mode.

  Note that this moving image code string conversion device can also be realized, for example, by using a general-purpose computer device as basic hardware.

  That is, the decoding unit, the encoding parameter conversion unit, the encoding parameter selection unit, and the encoding unit can be realized by causing a processor mounted on the computer device to execute a program. At this time, the moving image code string conversion device may be realized by installing the above program in a computer device in advance, or may be stored in a storage medium such as a CD-ROM or via the network. And may be realized by installing this program on a computer device as appropriate. B and C are realized by appropriately using a memory, a hard disk, or a storage medium such as a CD-R, a CD-RW, a DVD-RAM, a DVD-R, etc., which is built in or externally attached to the computer device. Can do.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

  In addition, the moving image code string conversion method and program according to one aspect of the present invention can be described as having any of the configurations or features exemplified below.

(1) A method of converting a first moving image code string obtained by compressing and coding an input signal of a jump order method into a second moving image code string,
A decoding step of decoding the first video code sequence as an input, generating a decoded image and extracting a coding parameter;
A coding parameter for a set of macroblocks arranged vertically in the second moving image code string is determined from the respective coding parameters of a plurality of corresponding macroblocks in the first moving image code string. Based on the encoding parameter, an encoding parameter conversion step for converting the encoding parameter included in the first moving image code string into an encoding parameter for the second moving image code string;
And a coding step for generating a second moving image code sequence using the coding parameter selected in the coding parameter conversion step.

(2) A method of converting a first moving image code string obtained by compressing and encoding an input signal of a jump order method into a second moving image code string,
A decoding step of decoding the first video code sequence as an input, generating a decoded image and extracting a coding parameter;
A part of encoding parameters for a set of macroblocks arranged vertically in the second moving image code string is determined from the respective encoding parameters of a plurality of corresponding macroblocks in the first moving image code string, Based on the determined encoding parameter, an encoding parameter conversion step for converting an encoding parameter included in the first moving image code sequence into an encoding parameter for the second moving image encoded sequence;
An encoding parameter generation step for generating an encoding parameter candidate from the decoded image or the converted second moving image encoded sequence;
An encoding parameter selection step for finally selecting a parameter to be used for encoding from the encoding parameter converted in the encoding parameter conversion step and the encoding parameter generated in the encoding parameter generation step;
And a coding step of generating a second moving image code sequence using the coding parameter selected in the coding parameter selection step.

(3) In the encoding parameter conversion step,
(1) or (2), wherein the coding structure of the macroblock in the first moving image code string is set to a field structure or a frame structure based on a prediction type or DCT type value of the macroblock Video code string conversion method.

(4) In the encoding parameter conversion step,
When the coding structure for the set of upper and lower macroblocks in the second video code sequence is the same as the number of macroblocks in the frame structure and the field structure based on the coding structure of the macroblock in the corresponding first video code sequence Has a field structure, and when the number of frame structures and the number of macroblocks in the field structure are different, the coding structure having the larger number is set as the coding structure of a set of upper and lower macroblocks ( The moving image code string conversion method according to 1) or (2).

(5) In the encoding parameter conversion step,
When the coding structure for the set of upper and lower macroblocks in the second video code sequence is the same as the number of macroblocks in the frame structure and the field structure based on the coding structure of the macroblock in the corresponding first video code sequence Has a field structure, and if the number of frame structures differs from the number of macroblocks in the field structure, the encoding structure with the higher evaluation value is used by using the evaluation value set based on the encoding parameter in the macroblock. Is set to an encoding structure of a set of upper and lower macroblocks. (1) or (2).

(6) In the encoding parameter conversion step,
When the coding structure of the set of upper and lower macroblocks in the second video code sequence and the coding structure of the corresponding macroblock in the first video code sequence are compared, and the coding structure is different, the first moving image (1) or (2) characterized in that a coding parameter included in a corresponding macroblock in an image code string is converted and generated in accordance with a coding structure of a set of upper and lower macroblocks in a second moving image code string. ) Moving image code string conversion method.

(7) In the encoding parameter generation step,
Coding parameters for intra-frame coding,
Encoding parameters for macroblock skip,
The moving image code string conversion method according to (2), wherein an encoding parameter relating to temporal direct prediction encoding is generated.

(8) In the encoding parameter generation step,
With respect to the encoding parameters related to interframe encoding acquired from the encoding parameter conversion step, if no motion vector is set, a motion vector is calculated by prediction from surrounding motion vectors, and the encoding parameters related to interframe encoding are used. The moving picture code string conversion method according to (2), characterized in that:

(9) In the encoding parameter selection step,
(2) The moving picture code string conversion method according to (2), wherein a prediction error signal is calculated and evaluation is performed using a sum of absolute differences of the prediction error signal when selecting an optimum parameter.

(10) In the encoding parameter selection step,
(2) The moving picture code string conversion method according to (2), wherein evaluation is performed using rate distortion characteristics when selecting an optimal parameter.

(11) In the encoding parameter selection step,
(2) The moving picture code string conversion method according to (2), wherein evaluation is performed using a generated code amount when selecting an optimum parameter.

  (12) The moving picture code string conversion method according to any one of (1) to (11), further including a resolution conversion unit.

(13) A program for converting, using a computer or a signal processor, a first moving image code string obtained by compression-coding an interlaced order input signal into a second moving image code string,
A decoding function that decodes the first moving image code string as an input, generates a decoded image, and extracts encoding parameters;
A coding parameter for a set of macroblocks arranged vertically in the second moving image code string is determined from the respective coding parameters of a plurality of corresponding macroblocks in the first moving image code string. Based on the encoding parameter, an encoding parameter conversion function that converts the encoding parameter included in the first moving image code sequence into an encoding parameter for the second moving image encoded sequence,
A moving image code string conversion program comprising: an encoding function that generates a second moving image code string using an encoding parameter selected by an encoding parameter conversion function.

(14) A program for converting, using a computer or a signal processor, a first moving image code string obtained by compression-coding an interlaced order input signal into a second moving image code string,
A decoding function that decodes the first moving image code string as an input, generates a decoded image, and extracts encoding parameters;
A part of encoding parameters for a set of macroblocks arranged vertically in the second moving image code string is determined from the respective encoding parameters of a plurality of corresponding macroblocks in the first moving image code string, Based on the determined encoding parameter, an encoding parameter conversion function for converting an encoding parameter included in the first moving image code sequence into an encoding parameter for the second moving image encoded sequence;
An encoding parameter generation function for generating an encoding parameter candidate from the decoded image or the converted second moving image encoded sequence;
An encoding parameter selection function for finally selecting a parameter to be used for encoding from the encoding parameter converted by the encoding parameter conversion function and the encoding parameter generated by the encoding parameter generation function;
A moving image code string conversion program comprising: an encoding function that generates a second moving image code string using an encoding parameter selected by an encoding parameter selection function.

1 is a block diagram of a video code string conversion device according to a first embodiment of the present invention. The flowchart of operation | movement of the moving image code sequence converter of the 1st Embodiment of this invention. The flowchart of the process of the encoding parameter conversion part in the 1st Embodiment of this invention. The figure which shows the relationship of the macroblock corresponding to the macroblock pair in the 1st Embodiment of this invention. The flowchart of the encoding structure setting process of the macroblock in the 1st Embodiment of this invention. 5 is a flowchart of processing for setting a coding structure of a macroblock pair according to the first embodiment of the present invention. The flowchart of the conversion process of the parameter of the corresponding macroblock in the 1st Embodiment of this invention. The figure which showed the example of conversion when the encoding structure of a macroblock and a macroblock pair differs in the 1st Embodiment of this invention (frame structure-> field structure). The figure which showed the conversion of the motion vector in the case of the frame structure in the 1st Embodiment of this invention, and a motion compensation block shape. The figure which showed the conversion of the motion vector in the case of the field structure in the 1st Embodiment of this invention, and a motion compensation block shape. The figure which showed the flow of the process of the encoding parameter selection part in the 1st Embodiment of this invention. FIG. 5 is a block diagram of a video code string conversion device according to a second embodiment of the present invention. The figure which shows the relationship of the macroblock corresponding to the macroblock pair in the 2nd Embodiment of this invention. The figure which shows the flow of the setting process of the encoding structure of the macroblock pair in the 2nd Embodiment of this invention. The block diagram of the moving image code sequence converter concerning the 3rd Embodiment of this invention. The figure which shows the example of conversion of the picture encoding type concerning the 3rd Embodiment of this invention. The figure which shows the example of conversion of the picture encoding type concerning the 3rd Embodiment of this invention. The figure which shows the example of the corresponding ratio of the macroblock corresponding to the macroblock pair in the 2nd Embodiment of this invention. The flow which shows the flow of a process of the macroblock pair in the 2nd Embodiment of this invention. The flow which shows the flow of a process of the macroblock pair in the 2nd Embodiment of this invention. The flow which shows the flow of a process of the macroblock pair in the 2nd Embodiment of this invention.

Explanation of symbols

100 ... 1st video code string
101 ... Decoded image signal
102 ... Coding parameters in the first video code string
103 ... 2nd video code string
104 ... Encoding parameter candidates
105 ... converted encoding parameters
106 ... determined suitable encoding parameters
107 ... Encoding information of the first moving image code string
110: Decoding unit for the first moving image code string
111 ... Coding parameter converter
112 ... Coding parameter selection section
113 ... Coding parameter generator
114 ... Encoding part of the second moving image code string
120 ・ ・ ・ Resolution converter

Claims (23)

An apparatus for converting a first code string obtained by compressing and encoding a video signal of an interlaced order scanning method in units of macroblocks into a second code string based on an encoding method in which compression encoding is performed in units of macroblocks,
A decoding unit that decodes the input first code string to obtain a decoded image and a first encoding parameter;
For each set of macroblocks related to the second code sequence adjacent in the vertical direction on the decoded image, the first of each of the plurality of corresponding macroblocks related to the first code sequence corresponding to the set on the image. An encoding parameter conversion unit that converts one encoding parameter to obtain a conversion encoding parameter;
An encoding parameter selection unit that selects the transform encoding parameter obtained for the set as a second encoding parameter of each macroblock of the set;
An encoding unit that compresses and encodes the decoded image using the second encoding parameter selected by the encoding parameter selection unit and generates the second code string;
An image code string conversion device. The image code string conversion device according to claim 1, further comprising:
An encoding parameter generation unit that generates encoding parameter candidates using the decoded image or the already generated second code string;
The encoding parameter selection unit selects a second encoding parameter of each macroblock of the set from the encoding parameter candidates and the transform encoding parameter.
The image code string converter according to claim 1. The first coding parameter is coding structure information indicating whether a coding structure of a macroblock is a field structure or a frame structure;
Prediction type information indicating the prediction type of the macroblock, and
DCT type information indicating the DCT type of the macroblock,
Including at least one of them,
The second encoding parameter and the transform encoding parameter are:
DCT type information indicating the DCT type of the macroblock,
Including
The encoding parameter converter is
Coding structure information of the corresponding macroblock;
Prediction type information of the corresponding macroblock; and
DCT type information of the corresponding macroblock,
Setting coding structure information of the transform coding parameter based on at least one of:
The image code string conversion device according to claim 1. The encoding parameter converter is
An extraction unit that extracts coding structure information of the first coding parameter of each of the corresponding macroblocks corresponding to each set;
For each set of coding structure information extracted by the extraction unit, a comparison unit that compares the number of corresponding macroblocks having a value indicating a field structure with the number of corresponding macroblocks having a value indicating a frame structure;
A value indicating a field structure is set in the coding structure information of each macroblock of the set having the same number of comparison results by the comparison unit, and each macroblock of the set having a large number of corresponding macroblocks in the field structure is set. A value indicating a field structure is set in the coding structure information, and a setting unit that sets a value indicating the frame structure in the coding structure information of each macroblock of the set having a large number of corresponding macroblocks in the frame structure;
The image code string converter according to claim 3, comprising: The encoding parameter converter is
An extraction unit that extracts coding structure information of the first coding parameter of each of the corresponding macroblocks corresponding to each set;
A determination unit that determines whether or not there is a corresponding macroblock having a value indicating a field structure for each set of coding structure information extracted by the extraction unit;
As a result of determination by the determination unit, if there is at least one corresponding macroblock having a value indicating the field structure, a value indicating the field structure is set, and if all the corresponding macroblocks have a value indicating the frame structure, a frame is set. A setting unit for setting a value indicating the structure;
The image code string converter according to claim 3, comprising: The encoding parameter converter is
An extraction unit that extracts coding structure information of the first coding parameter of each of the corresponding macroblocks corresponding to each set;
The corresponding macroblock corresponding to each set is classified by the value of the coding structure information of the first coding parameter, and the evaluation value for each classification is used using the first coding parameter of the corresponding macroblock belonging to the classification. An evaluation value calculation unit obtained by
A setting unit that sets coding structure information of each macroblock of the set according to the size of the evaluation value;
The image code string converter according to claim 3, comprising: The encoding parameter conversion unit further includes:
A comparison unit that compares the number of corresponding macroblocks having a value indicating a field structure with the number of corresponding macroblocks having a value indicating a frame structure for each set of coding structure information extracted by the extraction unit;
With
The setting unit of the encoding parameter conversion unit includes:
A value indicating a field structure is set in the coding structure information of each macroblock of the set having the same number of comparison results by the comparison unit, and the coding structure information of each macroblock in the set having a different number Set a value indicating the coding structure with the higher evaluation value,
The image code string converter according to claim 6. The evaluation value calculator in the encoding parameter converter is
For each of the corresponding macroblock group having a value indicating the field structure and the corresponding macroblock group having a value indicating the frame structure, the area corresponding to the set of macroblocks related to the second code string is calculated,
An evaluation value for each of the field structure and the frame structure is calculated based on the value of the corresponding area,
The image code string converter according to claim 6. In the coding parameter conversion unit, a coding structure setting unit for a set of macroblocks is:
7. The image code according to claim 6, wherein if the evaluation value of the field structure is equal to or higher than the evaluation value of the frame structure, the coding structure of the set of macroblocks is a field structure, and otherwise the frame structure is used. Column conversion device. In the coding parameter conversion unit, a coding structure setting unit for a set of macroblocks is:
If the value obtained by adjusting the evaluation value of the field structure with the weighting factor is equal to or higher than the evaluation value of the frame structure, the coding structure of the macroblock of the set is the field structure, and otherwise, the frame structure is used. And
The image code string converter according to claim 6. The setting unit of the encoding parameter conversion unit uniquely determines the weighting factor based on the relationship between the encoding information included in the first code string of the conversion source and the encoding information about the code string after conversion. Characterized by adaptive adjustment,
The image code string converter according to claim 10. The first coding parameter, the second coding parameter, and the transform coding parameter include coding structure information indicating whether a macroblock coding structure is a field structure or a frame structure,
The encoding parameter converter is
A coding structure information comparison unit that compares the coding structure information of each macroblock of the set and the coding structure information of each of the corresponding macroblocks;
A coding structure information conversion unit that converts the first coding parameters included in the corresponding macroblocks related to the sets having different coding structure information in accordance with the coding structures of the sets;
The image code string conversion device according to claim 1, comprising: The coding structure information conversion unit performs a conversion for scaling a motion vector included in the corresponding macroblock related to the set having different coding structure information according to a time axis.
The image code string converter according to claim 12. The encoding parameter generation unit includes:
Information about intra-frame coding,
Information about macroblock skip, and
Information about direct predictive coding,
Generate encoding parameter candidates including
The image code string converter according to claim 2. The first coding parameter includes prediction type information indicating whether each of the corresponding macroblocks is inter-frame coded or intra-frame coded in the first code string,
The encoding parameter generation unit includes:
Among the corresponding macroblocks, the intraframe-coded ones from the motion vector information of the macroblocks of the second code sequence that have already been compression-encoded around the macroblocks of the second code sequence, Means for calculating a motion vector prediction value for a macroblock;
Means for setting the calculated motion vector prediction value as an encoding parameter candidate of the macroblock for the corresponding macroblock that has been intraframe encoded;
The image code string conversion device according to claim 2, comprising: The encoding parameter selection unit
Selecting based on the sum of absolute differences of prediction error signals generated using each of the encoding parameter candidates and transform encoding parameters;
The image code string converter according to claim 2. The encoding parameter selection unit
Select based on rate distortion characteristics when compression encoding using each of the encoding parameter candidates and transform encoding parameters,
The image code string converter according to claim 2. The encoding parameter selection unit
Select based on the amount of generated code when compression encoding is performed using each of the encoding parameter candidate and the transform encoding parameter,
The image code string converter according to claim 2. It further comprises a resolution converter,
The image code string converter according to any one of claims 1 to 18. A method of converting a first code string obtained by compressing and encoding an interlaced sequential scanning video signal in units of macroblocks into a second code string based on an encoding method in which compression encoding is performed in units of macroblocks,
A decoding step of decoding the input first code string to obtain a decoded image and a first encoding parameter;
With respect to a set of macroblocks of the second code string adjacent in the vertical direction on the decoded image, the first encoding of each of the plurality of corresponding macroblocks of the first code string corresponding to the set on the image An encoding parameter conversion step for converting the parameter to obtain a conversion encoding parameter;
A coding parameter selection step of selecting the transform coding parameter obtained for the set as a second coding parameter of each macroblock of the set;
An encoding step of compressing and encoding the decoded image using the second encoding parameter selected by the encoding parameter selection unit to generate the second code string;
An image code string conversion method comprising: A coding parameter generation step of generating a coding parameter candidate using the decoded image or the already generated second code string;
In the encoding parameter selection step, a second encoding parameter of each macroblock of the set is selected from the encoding parameter candidates and the transform encoding parameter.
The image code string conversion method according to claim 20. A computer executes processing for converting a first code string obtained by compressing and encoding a video signal of an interlaced order scanning method in units of macro blocks into a second code string based on an encoding method in which compression encoding is performed in units of macro blocks. A program to
A decoding step of decoding the input first code string to obtain a decoded image and a first encoding parameter;
With respect to a set of macroblocks of the second code string adjacent in the vertical direction on the decoded image, the first encoding of each of the plurality of corresponding macroblocks of the first code string corresponding to the set on the image An encoding parameter conversion step for converting the parameter to obtain a conversion encoding parameter;
A coding parameter selection step of selecting the transform coding parameter obtained for the set as a second coding parameter of each macroblock of the set;
An encoding step of compressing and encoding the decoded image using the second encoding parameter selected by the encoding parameter selection unit to generate the second code string;
An image code string conversion program for executing A coding parameter generation step of generating a coding parameter candidate using the decoded image or the already generated second code string;
In the encoding parameter selection step, a second encoding parameter of each macroblock of the set is selected from the encoding parameter candidates and the transform encoding parameter.
The image code string conversion program according to claim 22.

Images