JP2784278B2 - Image scaling processor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
The present invention relates to an image scaling processing device that stores value data in a storage device, performs a scaling process as necessary, and outputs the data, and is suitable for application to a facsimile device, an image filing device, and the like.

[0002]

2. Description of the Related Art Conventionally, in a device such as a facsimile machine that handles binarized image data, in order to achieve mutual communication between devices having different paper sizes or images having different resolutions,
It is an indispensable technique to perform scaling processing such as enlargement and reduction on an input image. For example, in a memory communication mode in a facsimile machine, a document image read by a scanner is
In order to reduce the memory cost, binarization processing is performed, the data is temporarily stored as binarized data, and communication is started after a certain period of time has elapsed. Then, after the communication is started, the specification of the receiving apparatus is known for the first time, and if the paper size of the receiving apparatus is smaller than the size of the transmission original, the binary image data stored in the memory is subjected to reduction processing before transmission.

[0003] In such reduction processing, "thinning processing" for thinning out pixels at predetermined intervals is simple. With respect to the image subjected to the diffusion processing, the image quality is deteriorated such that the texture is disturbed.

[0004]

Therefore, a binary image is once converted into a multi-valued image having a desired pixel density, a scaling process is performed on the converted multi-valued image, and the converted multi-valued image is converted into a binary image again. There have been proposed various pixel density conversion devices that improve the image quality of a binary image after zooming (for example,
See JP-A-3-11879).

However, in these apparatuses, an interpolation method that simplifies the projection method is used for the binary / multi-value conversion unit, and a general error diffusion method is used for the multi-value / binary conversion unit. Therefore, moire occurs when an organized dither image is handled. Also,
In the projection method, an average density of original pixels projected onto pixels on the reduced image plane is obtained, and the obtained value is subjected to threshold processing to obtain the value of the reduced pixel (2
Value), a thin line (a line segment having a width of one dot) may be lost.

According to the present invention, when performing scaling processing of a binary image,
Suppress image quality deterioration such as moiré generated in the pseudo gradation image,
At the same time, it is an object of the present invention to provide an image scaling processing device that suppresses deterioration of image quality of a normal binary image such as a character or a line drawing.

[0007]

An image scaling processing apparatus according to the present invention separates a document image into a line drawing area and a picture area, and performs binarization processing suitable for each area; Attribute determining means for determining the attribute of the document image, memory means for storing the binary image data obtained from the binarizing processing means and attribute data obtained from the attribute determining means, and binary image data stored in the memory means. And scaling means for performing scaling processing suitable for the attribute of the document image based on the attribute data stored in the memory means.

[0008]

According to the present invention, the original image is adaptively subjected to different binarization processing between the line drawing area and the picture area by the binarization processing means, and the obtained binary image data is converted into attribute data representing the attribute of the original image. And stored in the memory means. When sending the binary image data stored in the memory means, if the scaling process is not required, the data is output as it is, and if the scaling process is required, the scaling is performed according to the attribute of the original image based on the attribute data. Process and output.

That is, if the original image is a line drawing image,
A scaling process is performed by a scaling method that does not cause the thin line to be lost or collapsed, for example, a logical sum method, and output. In addition, if the original image is a pattern image in which gradation is expressed, fine pixel arrangement has little meaning in the pattern image, and there is a problem in preserving the average density and texture within a certain range. After that, the image data is subjected to a scaling process by a pixel density conversion method such as a projection method, re-binarized by an error diffusion method or the like, and output. If the original image is an image in which a line drawing image and a picture image are mixed, edge pixels are detected from the binary image data, the edge pixels are left as they are, and the non-edge pixels are smoothed. The data is scaled, re-binarized by an error diffusion method or the like, and output.

[0010]

FIG. 1 is a block diagram showing an embodiment of the present invention. FIG. 1A shows a configuration for reading an original image, binarizing the original image, and storing it in a memory as image data. 1 (b) shows a configuration for scaling the image data stored in the memory and sending it out.

In FIG. 1A, a scanner 1 reads a target document image and has a resolution of about 400 dpi because of the necessity of reading characters. The image signals output from the scanner 1 are output in the order of raster scan, and are supplied to the line drawing processing unit 2, the picture processing unit 3, and the image area separation unit 4, respectively. The line image processing unit 2 is a means for performing a binary image encoding process by binarizing an input image signal by a predetermined threshold value, and the pattern processing unit 3 converts the image signal into a binary image by an area gradation method such as a dither method. Means for converting into pseudo halftone data based on the change in area.

The image area separating section 4 is means for separating a line drawing area and a picture area from an image signal. As a separation method, for example, a “line drawing separation method” (Japanese Patent Application No.
134054). In this method, an image signal is divided into blocks each having a predetermined number of pixels, and each block is compared with a predetermined pattern to detect the coincidence of the blocks.
The pattern is separated. The separation result in the image area separation unit 4 is supplied to the selection unit 5, and one of the binary image data processed by the line drawing processing unit 2 or the picture processing unit 3 is selected and stored in the memory 6.

The attribute determining section 7 determines the attribute of the document image using the separation result of the image area separating section 4 and stores the determination result in the memory 6 as attribute data. In this case, when the number of pixels determined to be a line image by the image area separation unit 4 is considerably small, it is determined that the image document is a document having only a picture pattern. It is determined that the image document is a line-only document, and that the other documents are mixed documents.

The binary image data stored in the memory 6 is supplied to the line drawing scaling section 8, picture scaling section 9 and line drawing / pattern mixed scaling section 10 shown in FIG. 1B. , And the attribute data are supplied to the control terminal of the selection unit 11.
The line drawing scaling section 8 is a means for performing scaling processing by a logical sum method in order to prevent thin character lines from disappearing due to scaling processing. In the OR method, a pixel value (2) is calculated by ORing pixels on a plurality of original images located near pixels on the reduced image plane.
Value), the thin line does not disappear. FIG. 2 shows the principle of the logical sum method. In the drawing, the mark ○ indicates a grid point of the original image, and the mark x indicates a grid corresponding to the grid point position of the scaled image on the original image. Here, the lattice interval of the original image is set to 1, and the scaling factor is set to k. Then, the grid interval of the scaled image is 1 / k. At this time, if at least one of the four pixels a, b, c, and d of the original image surrounding the pixel p of the scaled image is black, the scaled pixel p is set to black ("1").

The picture scaling unit 9 converts the binary image data into multi-valued image data, performs scaling processing, binarizes the data again, and outputs the result. FIG. 3 is a block diagram showing the configuration of the picture scaling unit 9. In the figure, input binary image data is converted by the binary / multi-level conversion unit 20 to multi-level image data having the same pixel density. The conversion into the multi-valued image data in the binary / multi-value conversion unit 20 is realized by a smoothing process using a 3 × 3 matrix size mask shown in FIG. This process is a low-pass filter process, and is a process for improving the image quality of the scaling process on the pseudo gradation image.
That is, since the pseudo-tone image is a pseudo-tone image expressed by the average density of the binary pixels within a certain range,
A fine shape within that range not only has little information but also causes image quality deterioration such as moire in the scaling process. Thus, by removing the high-frequency component in advance, the image quality of the pseudo-tone image subjected to the scaling process can be improved.

The scaling section 21 is means for converting image data output from the conversion section 20 as multi-value data into multi-value image data having different pixel densities. Here, this is performed by linear interpolation using three original pixels near the converted pixel. This linear interpolation method will be described with reference to FIG. As shown in FIG. 5 (a), the linear interpolation method uses four original pixels a, b, c, and 4 enclosing the variable pixel p projected on the original image plane.
Three pixels a, b, and d are selected from d. Then, the linear interpolation is performed by determining the contribution ratio of each density value of the selected three pixels to the scaled pixel p from the coordinates of the pixel p and the area ratio of the triangle formed by the vertices of the three pixels a, b, and d. I do. That is, as shown in FIG. 5B, the original pixels a, b,
Assuming that the density values of d are Ca, Cb, and Cd, the contribution ratio of the pixel a to the pixel p is determined by the ratio of the area Sc of the triangle at the diagonal position, and the contribution ratio of the pixel b is the area of the triangle at the diagonal position. The contribution ratio of the pixel d is determined by the ratio of the area Sa of the diagonal triangular position. Then, the density value Cp of the variable magnification pixel p is as follows: Cp = (Ca × Sc + Cb × Sb + Cd × Sa) / (Sa + Sb + Sc) This equation is equivalent to the following equation from the relative position (x, y) of the pixel p. Cp = (Cb-Ca) .x + (Cd-Cb) .y + Ca

Here, the pixel density after scaling is obtained by three-pixel interpolation. However, a projection method or a simplified four-pixel interpolation method (see Morita et al., “High-speed pixel density conversion method based on projection method”). ”, Journal of the Institute of Image Electronics Engineers of Japan, Vol. 11, No. 2
No., P72 to P83).

The density value of the pixel p after scaling obtained in this way.
The (multi-value) Cp is supplied to the multi-value / binary conversion means 22.
It is binarized. The binary conversion in the conversion means 22 is as follows.
This is performed by the error diffusion method shown in FIG. That is,
Error stored in the error buffer memory 30 so far.
Difference data ε_ijAnd the weighting factor α
_ijIs multiplied, and the adder 32 inputs the multi-valued data.
Ta x_ijIs added to FIG. 7 shows the weight coefficient α._ijShows an example
You.

The correction data x _ij ′ output from the adder 32 is compared with a threshold value th by a binarization circuit 33 to output binary data y _ij . In the arithmetic unit 34, the correction data x
The difference between _ij ′ and the binarized data y _ij is output as error data ε _ij and stored at the corresponding pixel position in the error buffer memory 30. By repeating this process, the binarization process of the input multi-valued image data _xij is executed.

Returning to FIG. 1B, the mixing scaling unit 10 determines whether each pixel of the binary image data is an edge pixel, and if the pixel is an edge pixel, scales the binary image data as it is. If the pixel is a non-edge pixel, it is converted into multi-valued image data, scaled, and then binarized again and output. FIG. 8 shows the configuration of the variable power unit 10 for mixing. The scaling unit 10 includes an edge detection unit 23 that determines whether each pixel of the binary image data supplied from the memory 6 is an edge pixel.
And input 2 based on the determination result of the edge detection unit 23.
3 except that a selection unit 24 for selecting one of the value image data or the multi-valued image data output from the binary / multi-value conversion conversion unit 20 is provided. It has the same configuration. Therefore, the selecting section 24 outputs image data in which only the non-edge portion of the input binary image data is converted into multi-valued data.

The edge detection in the edge detection unit 23 is as follows.
The arrangement of the pixels around the target pixel is represented by eight types of patterns shown in FIG. 9, that is, a white (or black) pixel column having three pixels arranged vertically or horizontally and a black (or white) pixel column having a length of three pixels Among the eight types of adjacent patterns, pixels that match at least one pattern (pixels marked with a circle in the figure) are detected as edge pixels. An edge pixel means a vertical or horizontal line segment of a character or the like.

The image data subjected to the scaling processing by the line drawing scaling section 8, the picture scaling section 9, and the mixture scaling section 10 are
Any one of the output image data is selected by the selection unit 11 based on the attribute data, and output as binary image data after scaling. In the above-described embodiment, it is simply determined whether or not the original is only a line image, and the original with only the line image is scaled by the logical sum method. The magnification may be changed in the same manner.

The determination of the document attribute may be set manually instead of automatically.

[0024]

According to the present invention, the image attribute is determined at the time of reading a document image, so that attribute determination can be performed with high accuracy. In addition, based on the determination result, a scaling processing method suitable for each image attribute is performed. Since the switching is performed, high-quality scaling processing can be performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a diagram for explaining the principle of the logical sum method.

FIG. 3 is a block diagram of a picture scaling unit.

FIG. 4 is a diagram showing a binary / multi-value conversion mask.

FIG. 5 is a diagram for explaining linear interpolation.

FIG. 6 is a block diagram of a multivalue / binary conversion unit.

FIG. 7 is a diagram illustrating an example of a weight coefficient.

FIG. 8 is a block diagram of a variable power unit for mixing.

FIG. 9 is a diagram illustrating an example of an edge pixel detection pattern.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Scanner 2 Line drawing processing part 3 Picture processing part 4 Image area separation part 5 Selection part 6 Memory 7 Attribute judgment part 8 Line drawing scaling part 9 Picture scaling part 10 Mixing scaling part 11 Selection part

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06T 3/40 G06T 5/00 H04N 1/393

Claims (5)

(57) [Claims] 1. A binarization processing unit for separating a document image into a line drawing area and a picture area, and performing a binarization process suitable for each area; an attribute determination unit for determining an attribute of the document image; Memory means for storing binary image data obtained from the binarization processing means and attribute data obtained from the attribute determination means; and the attribute storing the binary image data stored in the memory means in the memory means An image scaling processing device, comprising: a scaling unit that performs scaling processing suitable for the attribute of the document image based on data. 2. The magnification processing unit according to claim 1, wherein the magnification processing unit includes a line drawing magnification unit that performs magnification processing on a line image such as a character and a line drawing; Image resizing means for performing a resizing process on a mixed image in which line images and picture images are mixed, and selecting one of the resizing means based on the attribute data. apparatus. 3. The line drawing scaling unit according to claim 2, wherein the line drawing scaling unit uses a logical sum method of determining a pixel value by a logical sum of pixels on a plurality of original images located near pixels on the reduced image plane. An image scaling processing device, characterized in that it is means. 4. The image magnification changing means according to claim 2, wherein said pattern scaling means performs a smoothing process on the input binary image data to convert the input binary image data into multi-valued data, and calculates a density after scaling by a projection method or an interpolation method. An image scaling processing device, characterized in that it is means for re-binarizing by interpolating by a quantization error accompanying binarization of a peripheral pixel that has already been binarized. 5. The variable magnification means for mixing according to claim 2, further comprising means for detecting an edge pixel of the input binary image data, and smoothing the edge pixel as it is and a non-edge pixel as it is. An image scaling processing device, which performs a scaling process by a projection method or an interpolation method, and re-binarizes by an error diffusion method.