JP5754959B2 - Image processing apparatus, image processing method, and program - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and a program, and more particularly to a technique suitable for use in tone correction.

  Recent printers are equipped with various gradation correction means so that even underexposed images can be printed with appropriate brightness. There are various methods for correcting the gradation, and one of them is a correction method using histogram information. In this correction method, a luminance histogram or RGB histogram is acquired from the image, and the entire image is adjusted to a tone curve so that a gradation value that has a cumulative cumulative frequency equal to or higher than a certain ratio becomes a predetermined gradation value. Correction is performed by processing or gain processing.

  On the other hand, the image includes various subjects, but not all subjects are information necessary for the correction process. It is desirable that the photographer shoots to capture the so-called main subject, and that the main subject is corrected with an appropriate brightness.

  However, since the histogram used in the above method has integrated information of brightness and RGB values, information on all subjects included in the image is handled with the same importance. Therefore, if the main subject is underexposed and the other subject occupies much of the highlight area, the main subject information is buried in the other subject's information, and the main subject is underexposed from the histogram. It becomes impossible to judge. As a result, the exposure of the main subject cannot be corrected appropriately.

  In order to solve such a problem, a method is disclosed in which an area that does not need to be considered in the correction process is determined in an image, and correction is performed using only a histogram acquired from other areas. For example, Patent Document 1 discloses a method in which an image is divided into a plurality of partial areas, and gradation is corrected based on a local histogram acquired from these areas.

  The technique described in Patent Document 1 divides an image into a plurality of areas, and among these areas, an area having a point with the lowest luminance, an area having a point with the highest luminance, and the variance of the histogram. The low region is a region that must be considered in the correction process. Then, an intermediate histogram is created by applying predetermined weights to the histograms of these regions, and a tone curve used for tone correction processing is calculated from the intermediate histogram. That is, a region that does not match any of these three conditions is determined as a region that does not need to be considered in the correction process, and is not reflected in the intermediate histogram.

Japanese Patent No. 4398595

  However, in the technique described in Patent Document 1 described above, even if the region does not match any of the above three conditions, not all of the regions need not be considered in the correction process.

  For example, in the image as shown in FIG. 3A, the sky is shown at the top and the main subject is shown at the center. FIG. 4A is a diagram showing an RGB histogram of the image shown in FIG. FIG. 4B is a diagram showing an empty histogram in the image shown in FIG. 3A, and FIG. 4C is a histogram of the subject in the image shown in FIG. FIG. Comparing the histogram shown in FIG. 4 (a) with the histogram shown in FIG. 4 (b), it can be seen that the sky occupies most of the highlight portion. From the histogram shown in FIG. 4 (c), the subject is slightly underexposed. It can be seen that it is.

  In order to optimally correct an image having many highlight portions in this manner, it is desirable to brightly correct the main subject that is underexposed. However, when the above three regions are selected, the main subject may not be selected as a region to be considered in the correction process depending on the variance of the RGB histogram of the main subject. Therefore, in a scene in which other subjects occupy more highlight portions than the main subject, the main subject is not appropriately subjected to gradation correction.

  The present invention has been made in view of the above-described problems, and it is an object of the present invention to appropriately perform gradation correction on a main subject in a scene in which a subject other than the main subject occupies as a highlight portion.

The image processing apparatus of the present invention calculates a first calculation means for calculating the distribution information of the signal value for each divided portion region of the input image the distribution information of the signal value of the input image A second value calculating unit, a first value that is set based on a signal value in which the cumulative frequency from the top in the distribution information calculated by the first calculating unit is a predetermined value; and the second calculating unit. The comparison means for comparing the second value which is the median value in the distribution information calculated by the above, and the second value is determined to be larger than the first value by the comparison means from the input image Correction means for performing gradation correction of the entire input image using an image area excluding the partial area.

  According to the present invention, it is possible to appropriately perform gradation correction even in a scene in which many highlight portions other than the main subject occupy.

It is a block diagram which shows the function structural example of the image processing apparatus which concerns on embodiment. It is a flowchart which shows an example of the process sequence of the gradation correction | amendment in embodiment. It is a figure which shows an example of the image in which the sky and the main subject are reflected. It is a figure which shows an example of the RGB histogram of the whole and the part. It is a figure which shows an example of the RGB histogram and tone curve after removing the unnecessary area | region.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram illustrating a functional configuration example of the image processing apparatus according to the present embodiment.
In FIG. 1, an image input unit 102 inputs image data to be subjected to gradation correction processing from an external storage device such as a CD-ROM or a memory card. Here, various file formats are conceivable as the file format of the input image data. In this embodiment, the Jpeg format used for printing is assumed.

  The image format conversion processing unit 103 converts the image data input by the image input unit 102 into RGB format. Specifically, the Jpeg format is expanded to the YCbCr format, and further converted to the RGB format. The tone correction unit 104 calculates a correction amount necessary for tone correction based on the RGB format image data acquired by the image format conversion processing unit 103, and performs tone correction processing.

  The image format inverse conversion processing unit 105 inversely converts RGB format image data into the original image format. Specifically, conversion from RGB format to YCbCr format is performed, and further, compression from YCbCr format to Jpeg format is performed. The image output unit 106 outputs the Jpeg format image data compressed by the image format inverse conversion processing unit 105 to the outside. Here, the output destination is assumed to be overwritten and saved in the external storage device in which the original image data is stored.

  When the image data input to the image input unit 102 is RGB format image data, the gradation correction unit 104 directly inputs the image data, and the image output unit 106 does not pass through the image format reverse conversion processing unit 105. Directly output the corrected image data.

FIG. 2 is a flowchart illustrating an example of a tone correction processing procedure performed by the tone correction unit 104 according to this embodiment. Hereinafter, an example in which gradation correction is performed on an image as shown in FIG.
First, in S 201, RGB format image data is input from the image input unit 102 or the image format conversion processing unit 103. In the present embodiment, it is assumed that RGB format image data relating to the image shown in FIG.

  Next, in S202, the specific direction (sky direction) of the image data is determined. Here, the sky direction is the sky direction that the digital camera faces when an image is taken. Various techniques are disclosed in this determination processing, but generally, such information is included in metadata of image data, and this information is used in the present embodiment. In the case of the image shown in FIG. 3A, the upper part is the sky direction.

  Generally, the sky is brighter than the main subject, and as a result, it occupies a lot of highlight portions of the histogram. Therefore, in the present embodiment, the highlight information due to the sky is treated as unnecessary information when calculating the gradation correction amount.

Next, in S203, the tone correction unit 104 functions as first calculating means, by sampling from the input image data, calculates an RGB histogram H _all. Then, in S204, to calculate a first highlight point RGB _hi1 the RGB histogram H _all a second highlight point RGB _hi2.

In the present embodiment, RGB histogram H _all uses RGB values to be 99% cumulative frequency as a first highlight point RGB _hi1. Further, RGB histogram H _all uses an RGB value of 90% cumulative frequency as a second highlight point RGB _hi2. Then, in S205, using these highlight points, it is determined whether or not the highlight ratio in the image is smaller than the threshold by the following equation (1).

Here, the number of gradations existing between the first highlight point RGB _hi1 (accumulation frequency 99%) and the second highlight point RGB _hi2 (accumulation frequency 90%) is defined as the highlight gradation number. To do. If the difference between the two highlight points is equal to or greater than a predetermined value (threshold Th _RGB1 or greater), it is determined that the highlight ratio in the image is small. Here, although various values can be set as the threshold Th _RGB1, it is assumed to be 15 in this embodiment. As a result of the determination in S205, if the difference is _greater than or _equal to the threshold Th _RGB1 , the processing in S206 to S213 described later is unnecessary, and the process directly proceeds to S214.

On the other hand, if the result of determination in S205 is that the difference is less than the threshold Th _RGB1 , the process proceeds to S206. In S206, in order to discriminate an image area that is considered unnecessary when calculating the gradation correction amount, as shown in FIG. 3B, the image is divided into a plurality of areas (for example, 10) in the sky direction. The recognition number n is added in order from the top. In S207 and subsequent steps, it is determined whether or not each of the partial areas thus divided is an area unnecessary for calculating the gradation correction amount.

In S207, first, the recognition number n is set to 1. In step S208, the gradation correction unit 104 functions as a second calculation unit, and calculates a local RGB histogram H [n] for that region. FIG. 4D shows the local RGB histogram H [1]. Next, in S209, the median RGB _middle [n] of the local RGB histogram H [n] is calculated. The median value RGB _middle [n] indicates a representative gradation value of the partial area of the identification number n. Next, in S210, in order to determine whether or not this region in the image belongs to the highlight, it is determined whether or not the following expression (2) is satisfied.

Here, the coefficient α is a parameter for adjustment, and 0.90 that gives the best result is used in this embodiment. In this way, the value obtained by multiplying the second highlight point RGB _hi2 by the coefficient α as the first value and the median value RGB _middle [n] of the local RGB histogram H [n] as the second value are compared. . Further, the reason why the median value is used instead of the average value as the amount indicating the representative gradation value of the partial area is as follows. This is because when the histogram is biased toward a low gradation, if the average value is used, the representative gradation value of the partial area is estimated to be low even if the highlight occupies a large amount.

If the expression (2) is satisfied as a result of the determination in S210, it means that the median value RGB _middle [n] belongs to a region higher than the second highlight point RGB _hi2 , and the recognition number n in the image This means that the partial area of belongs to the highlight. Thus, when the highlight portion belongs to the sky direction of the image, it affects the calculation of the gradation correction amount and is not used for calculating the gradation correction amount because the main subject is not corrected appropriately. . Therefore, in this case, the process proceeds to S211. On the other hand, as a result of the determination in S210, if the expression (2) is not satisfied, it does not belong to the highlight, so it can be determined that the subsequent partial areas also do not belong to the highlight. Accordingly, the process proceeds to S213.

  In S211, the recognition number n is incremented by one. In S212, it is determined whether or not the recognition number n is 4 or more. If the result of this determination is less than 4, the process returns to S208 to determine the next area. On the other hand, if it is 4 or more, the process proceeds to the next S212. In this embodiment, the identification number criterion for determining whether or not it belongs to the highlight point is set to n = 4.

FIGS. 4E to 4G are diagrams showing local histograms H [n] for n = 2 to 4, respectively. As shown in FIG. 4D to FIG. _4G , RGB distribution [3] satisfies Expression (2) in each distribution.

Therefore, in the next S213, as shown in FIG. 3B, the partial areas up to the recognition numbers 4 to 10 excluding the partial areas of the recognition numbers 1 to 3 determined to be unnecessary are necessary for gradation correction. The region is then sampled again to obtain the RGB histogram _Hall . FIG. 5A is a diagram showing an RGB histogram H _all of an area necessary for tone correction. As shown in FIG. 5A, since the sky highlight portion is deleted, the main subject information is included more than the RGB histogram calculated from the entire image shown in FIG. Thus, more appropriate correction is possible.

Next, in S214, it computes the tone curve used in the original to the tone correction processing an RGB histogram H _all calculated again S213. Various methods for calculating the tone curve are disclosed. In the present embodiment, as in the third highlight point RGB _{Hi 3} at a cumulative frequency of 99% of the RGB histogram H _all is more TargetRGB, using the tone curve as shown in the following equation (3).

  Here, RGBin is the number of input gradations (number of gradations before gradation correction), and RGBout is the number of output gradations (number of gradations after gradation correction). In this embodiment, 230 is used as TargetRGB, and the calculated tone curve is shown in FIG. Next, in S215, tone correction is performed on RGB of the entire image using the tone curve calculated in S214.

  FIG. 5C is a diagram showing an image histogram that has been subjected to gradation correction by these processes. The histogram indicated by the dotted line is the result of correction by a conventional method. On the other hand, in the present embodiment, when calculating the gradation correction amount, a histogram having a lot of information on the main subject is used, so that the exposure of the subject is corrected more appropriately. Finally, in S216, the corrected image data is output to the image format inverse conversion processing unit 105 or the image output unit 106.

  As described above, according to the present embodiment, since the highlight portion of the image is not used for gradation correction, gradation correction can be appropriately performed even in a scene where a lot of highlights other than the main subject are occupied. .

(Other embodiments)
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

  104 Tone correction unit

Claims (7)

First calculation means for calculating distribution information of signal values of input images;
A second calculating means for calculating distribution information of the signal value for each divided portion region of the input image,
In the distribution information calculated by the first calculation means, a first value set based on a signal value whose cumulative frequency from the top is a predetermined value, and distribution information calculated by the second calculation means comparing means for comparing the second value to be the median,
Tone correction of the entire input image is performed using an image area obtained by excluding a partial area where the second value is determined to be larger than the first value from the input image by the comparison unit. An image processing apparatus comprising correction means. The image processing apparatus according to claim 1, wherein the partial region is a region divided in the vertical direction of the input image. The comparison means performs a comparison in order from the upper partial area of the input image, and when it is determined that there is a partial area in which the second value is not greater than the first value, a lower portion than the partial area. The image processing apparatus according to claim 2, wherein the partial areas are not compared. The image processing apparatus according to claim 2, wherein the comparison unit does not compare a partial area below a predetermined partial area of the input image. 5. The image according to claim 2 , wherein the vertical direction of the input image is a direction determined based on metadata added to the input image. 6. Processing equipment. A first calculation step of calculating distribution information of signal values of the input image;
A second calculation step of calculating a distribution information of the signal value for each divided portion region of the input image,
In the distribution information calculated in the first calculation step , a first value set based on a signal value whose cumulative frequency from the top is a predetermined value in the distribution information calculated in the first calculation step , and the distribution information calculated in the second calculation step A comparison step of comparing the second value to be the median value ;
Using the image area excluding the partial region determined as the second value in said comparing step from said input image is greater than said first value, performing the input image overall tone correction An image processing method comprising: a correction step. A first calculation step of calculating distribution information of signal values of the input image;
A second calculation step of calculating a distribution information of the signal value for each divided portion region of the input image,
In the distribution information calculated in the first calculation step , a first value set based on a signal value whose cumulative frequency from the top is a predetermined value in the distribution information calculated in the first calculation step , and the distribution information calculated in the second calculation step A comparison step of comparing the second value to be the median value ;
Using the image area excluding the partial region determined as the second value in said comparing step from said input image is greater than said first value, performing the input image overall tone correction A program for causing a computer to execute a correction process.

Images