JPH0679337B2 - Image processing method - Google Patents

Description

The present invention relates to an image processing method for converting an image.

[Prior art]

Conventionally, as one of the image processing methods of this kind, there are known an electronic plate making technique in the printing industry and a professional laboratory technique applying computer photographic image processing in the photographic laboratory industry.

For example, when a reproduced image is obtained by photoelectrically scanning an image original with a high-precision scanner such as a layout scanner or a laser color printer, a signal processing unit is provided in the middle of the process to correct the density expression (γ correction) for the input density signal. ), Gradation setting,
The following effects are obtained by performing processing such as color correction and clipping composition.

Restoration of fading color film.

Adjust the gradation of highlights and shades to exaggerate the color expression.

Relief of equipment failure, shooting mistake, development mistake, etc.

Expressing creative images, expanding the image area, and creating new designs.

When such special processing is executed, a digital image such as a density signal or a luminance signal of the film original obtained by photoelectrically scanning the film original with a color scanner, a high-precision scanner or a color image pickup tube, a color image pickup plate (for example, CCD), etc. Processing the signal.

Conventionally, this kind of special effect includes mosaic processing arranged regularly, posterization processing for changing the γ curve into an unrealistic one, solarization processing, and the like.

The conventional regularly arranged mosaic processing has the following drawbacks, and an improved processing method is desired for expressing a creative image.

The arrangement of rectangular blocks is too regular.

Since all the pixels in the block have the same value, the information where the original image requires a fine expression (where there are many high frequency components) is lost.

Further, for example, in a low frequency region of an image such as a portrait image, the background and the person have a mosaic of the same shape, which makes the impression of the subject light.

Furthermore, the conventional processing method does not sufficiently reproduce the touch (using a brush) of a paint brush as seen in oil paintings, etc., and there is a demand for image processing that detects the edge directionality of an image and makes the best use of this.

[Object of the Invention]

The present invention is. By eliminating the drawbacks of the above-mentioned conventional example, we provide a new image processing method to the few conventional special effect processing, express a creative image without being influenced by the sense of the creator, expand the image area, new It is very useful for creating designs, and contributes to the advertising and design industries dealing with photographic images, for example, and makes it easy for general consumers to use special photographic techniques that are difficult or impossible with conventional photographic technology. The purpose of the present invention is to provide an image processing method that can be incorporated into the image processing method and can contribute to the industrial development of the photographic industry, the photo laboratory industry, and the like.

〔Example〕

FIG. 1 is a block diagram showing an example of the flow of a density signal when the present invention is applied to a color scanner. An input signal obtained by photoelectrically scanning a film original loaded on the input drum 1 is converted into a density signal by the logarithmic conversion circuit 2 and then input to the input signal processing unit 4 via the AD converter 3. , The density value of each film is converted according to the type of the film, and the density value is converted so as to have a fixed relationship with the exposure amount of the film. The film characteristic curve differs depending on whether the film is a negative film, a positive film, or a reversal film.
This is because each color is different. After that, the image is input to the image processing unit 5, which should be the heart of the system, and digital image processing is performed. After that, the density signal output from the image processing unit 5 is processed by the output signal processing unit 6 to be converted into a laser light amount control signal, input to the modulator 8 via the DA converter 7, and output from the light source 9. The output laser 10 is caused to reproduce a desired image by modulating the generated laser light.

FIG. 2 is a block diagram showing a part of the image processing unit 5 in more detail. The CPU 11 is a control processor and has all the real rights of the image processing unit 5.
Image processing is controlled by the memory. The parameter controller 13 controls the arithmetic unit 14, the parameter memory 15, and the parameter setting I / O 16 to perform initialization, setting, comparison, etc. of parameters necessary for processing. Processor 18 and CP
An image controller 17 connects between U11, and the processor 18 operates according to a command from the CPU 11. The processor 18 is a core part of the image processing unit 5, and outputs image data from any of the image memories 20 to 26 selected according to the instruction of the CPU 11 and the 16-bit integrated image memory 27 and the I / O 36 for image data. Any image memory 20 that receives and calculates the result of calculation is selected
Integrated image memory 27 for 26 and 16 bits and I / O for image data
Send to O36. In addition, the data can be output to the flag map memory 19, which is a memory dedicated to the carrier, in order to carry the carrier to a necessary address by calculation. Image memory 20-2
The cumulative image memory 27 for 6 and 16 bits is CPU BUS, VIDEO B
Since it is connected to any of the US, any image memory 20 to 26, 16 bit total image memory 27 from CPU 11
It is also possible to read and write to and to perform real-time arithmetic between arbitrary memories. A high-speed RAM called Lucupup table 28-34 is connected to the exit of the image memory. This RAM has a structure of 256 x 6 bits and has 8 RAM address lines (0 to 256 addresses can be specified = 0 to 255
The gradation can be specified) is directly connected to the output of each image memory, and 8 data lines are connected to the VIDEO BUS. The contents of RAM can be freely read and written by the CPU 11.
The image data I / O 36 is an image input / output interface, which inputs image data from the input signal processing unit 4 of FIG. 1 and outputs image data to the output signal processing unit 6.

FIG. 3 is a graph showing the states of the lookup tables 28 to 34, in which the horizontal axis shows the input gradation and the vertical axis shows the output gradation. FIG. 3 (a) shows a standard state in which a value 0 is written in address 0, a value 1 is written in address 1, ... 255 in address 255, and the contents of input and output are the same. . Further, FIG. 3 (b) shows that a value of 255 is written in the 0th address, a value of 254 is written in the 1st address, ..., A value of 0 is written in the 255th address, and the input-output is an image in which the density data is inverted. .

The digital image processing method performed by the image processing unit 5 will be described below in detail with reference to the flow chart of FIG. Here, the density data of the digital image is shown in the form of ai (m, n) for the sake of explanation. However, i is RGB and indicates R component image data, G component image data, and B component image data of the original image, respectively. Each pixel is composed of 8 bits for each component, that is, data capable of expressing 256 gradations. The value of the data with the highest density (darkest data) is φ, and the data with the lowest density (the brightest data) Value 255
And

Description of Step 1) In response to an instruction from the CPU 11, the processor 18 stores the R component image data of the original image in the image memory (1) 20 from the input signal processing unit 4 via the image data I / O 36. Then, the G component image data and the B component image data of the original image are stored in the image memories (2), (3), 21, 22 respectively. At this time, the rack up tables (1) to (7) and 28 to 34 are shown in FIG. 3 (a).
Is the standard condition.

Description of Step 2) Reference image data to be referred to when determining the directionality of the image edge portion is created in the following step and stored in the image memory (4) 23. In this embodiment, R component image data in the image memory (1) and G in the image memory (2) are used.
The component image data and the B component original image data in the image memory (3) are added together at a ratio of R: G: B = 3: 6: 1, and monochrome reference image data is created as image data with near visual sensitivity. However, it is also possible to use only the G component image data having the highest RGB ratio as the visual sensitivity as the reference image data. Processor 18
Function resets the 16-bit image memory (8). Next, the image memory (1) is added to the image memory (8) three times, and the data of the image memory (2) and the data of the image memory (3) are respectively added to the data 6 times.
Enjoy once, once. Image memory (1),
The data of (2) and (3) are ai (m, n) i = RGB.
And then, the data of the image memory _{(8) a 8 (m,} n) to represent the _{a 8 (m, n) =} 3 × a R (m, n) + 6 × a G (m, n) + a B
(M, n). Next, the data in the image memory (8) is divided by 10 and stored in the image memory (4).

That is, the data a ₄ (m, n) in the image memory (4) is a ₄
( _M , n) = (3 × a _R (m, n) + 6 × a _G (m, n) + a _B (m, n
n)) / 10.

Explanation of Step 3) Image data stored in the image memories (1), (2) and (3) are converted into image memories (6), (7) and (8).
The image data of the image memory (1), (2), and (3) are all set to 0, and the image data of the image memory (4) is copied to the image memory (2). Further, all the bits on the flag map memory 19 are set to 1. Then, using the digitizer (not shown) attached to the parameter setting I / O 16 for this image memory (2), a closed curve surrounding the main subject area is written. The inside of this closed curve is painted out, this portion is defined as the first area (main subject area), and (i, j on the flag map memory 19 corresponding to the pixel position (i, j) existing in this area). Write 0 to j). Further, the portion having 1 on the flag map memory 19 is defined as the second area (background area). After this, the image memory (6),
The data of (7) and (8) are stored in the image memory (1),
Return to (2) and (3).

Description of Step 4) A more detailed flow chart of Step 4 is shown in FIG. First, in step 4-1, the reference image data (which is on the image memory (4)) is transferred to the image memory (5),
Copy to (6) and (7).

Next, in step 4-2, these image memories (4),
Images of (5), (6), and (7) a _K (i, j) (K = 4,5,6,
7) is sequentially subjected to a known convolution process using direction detection operators each having a window size of 3 × 3 of OP1 to OP4 shown in FIG.
This processing result is again used as the original image memory (4),
Store in (5), (6) and (7). Here, OP1 is an operator for edge detection in the direction from upper right to lower left, and
2 is a horizontal direction, OP3 is an upper edge to lower right diagonal direction, and PO4 is a vertical edge detection operator.

Next, step 4-3 will be described. The values of the image memories (4), (5), (6) and (7) are compared for each pixel position (i, j), and the image memory number having the maximum value is searched for. For example, when a ₄ (i, j) in the image memory (4) is larger than other a _K (i, j) (K = 5,6,7), the 3 × 3 part centered on this position In the area, the edge component in the diagonal direction of 45 ° from the upper right to the lower left is large, and 1 is written to the position of (i, j) of the image memory (4), and its value a ₄
Write (i, j) to the position (i, j) in the image memory (5).

Similarly, a ₆ (i, j) in the image memory (6) is a
_When it is larger than _K (i, j) (K = 4,5,7), it is diagonally 45 from the upper left to the lower right in the 3 × 3 partial area which is the center of this position.
Since the edge component in the direction of ° is large, 3 is written to the position (i, j) of the image memory (4) and the value a ₆ (i, j
Write j) to the location (i, j) of the image memory (5). By carrying out this processing for all (i, j), the direction edges 1 to 4 are stored in the image memory (4), and the value in the direction of the largest edge component is stored in the image memory (5). .

Note that, in general, when the image memory number having the maximum value in (i, j) is N, N-3 is written in (i, j) of the image memory (4) and the maximum value is stored in the image. It will be written in (i, j) of the memory (5).

Explanation of Step 5) Here, the parameters necessary for the subsequent calculations are set. In the present embodiment, a parameter representing the size of the basic pixel block and a parameter representing the shape of the pixel block are input from a parameter setting I / O 16 using a keyboard (not shown). In the following example, the parameter indicating the size of the basic pixel block is given in the form of mo and no on one side of the circumscribed quadrangle of that shape, and the shape of the pixel block is determined by the direction detection operators OP1 to OP4 in FIG. Correspondingly give 4 kinds.

If the pixel block size is 7 × 7, the shape of the pixel block corresponding to PO1 is as shown in FIG. 8, and the block shape shows the directionality from upper right to lower left. Here, the pixel portion to which 1 is given indicates a pixel to be processed as a mosaic, and the portion 0 is a pixel to be not processed.

Here, a determination threshold TH for detecting a more meaningful directionality is given from the keyboard. This threshold TH is the maximum directivity value a _K (i, j) stored in the image memory (5).
It is used for processing. Further, the pixel block shape corresponding to the case of being less than this threshold value is inputted in the same manner as in FIG. The image block shape is preferably a shape that does not show much directionality, and an example thereof is shown in FIG.

The above parameters are set for each area of the reference image divided in step 3.

Furthermore, input the number of mosaics NSTOP to be generated below. For this, the area ratio P of the mosaic may be input and calculated from the following formula.

However, N _X and N _Y are the number of pixels in the X and Y directions of the input image.

Description of Step 6) Using the threshold value TH input in Step 5, it is checked whether or not each pixel value of the image memory (5) in which the maximum directional value is stored is larger than this. If the value of the image memory (5) at (i, j) is smaller than the threshold value T _H , the value of 0 is written to the position (i, j) of the image memory (4). In other words, since no significant directionality can be detected at this position, the pixel block shape shown in FIG. 9 input in step 5 is used to perform unconditional conversion processing.

In the present embodiment, since the reference image is divided into two regions, the flag map memory 19 is used, and when the flag map at (i, j) is 0, that is, the main subject region, TH1 is used as the threshold value, and (i, j When the flag map in j) is 1, that is, in the background area, TH2 is used as the threshold value. This allows
It is possible to change the ratio of the mosaic processing considering the directionality in the main subject area and the background area.

Explanation of Step 7) A more detailed flow chart of Step 7 is shown in FIG.

Explanation of Step 7-1) A random position memory and an image memory (5) for storing randomly generated coordinate positions are prepared, and all values are set to 0.

Explanation of Step 7-2) The counter count for counting the number of random coordinates is set to 1.

Explanation of Step 7-3) This is a step for obtaining a coordinate position by generating a random number.

Explanation of Step 7-4) The random position memory corresponding to the generated random coordinate positions i _R , T _R is set to 1.

Explanation of step 7-5 Count up the counter count, which counts the number of random coordinates, by one.

Description of Step 7-6) Steps 7-3 to 7-5 are repeated until the value of the counter count for counting the number of random coordinates reaches the required number NSTOP, and when NSTOP is reached, the process proceeds to the next step.

Explanation of step 8) Direction data Idirection is set to 0. As a result, in the following processing, the processing is performed on the portion that is equal to or greater than the threshold value TH in step 6 described above.

Explanation of step 9) From this point onward, since the values of the pixels are combined while being shifted, all the values of the shift memory image memories (6), (7) and (8) for this purpose are set to 0. To do.

Explanation of Step 10) Random position memory The image memory (5) has a position of 1 (that is, a random coordinate position), and the direction image data image memory (4) is Idirection.
Select the coordinate position that is equal to (currently 0).
As a result, there is no directivity, that is, the maximum directivity is
A random coordinate (i, j) value smaller than TH is selected. Then, for this selected coordinate position (i, j), the value of each input image data a _K (i, j) (K = 1,2,3) is assigned to each shift memory a _K (i, j). ) (K = 6,7,8).

Description of Step 11) The input mosaic pixel pattern whose direction data corresponds to Idirection (currently 0) is read out from the file created in the previous Step 5. FIG. 9 shows an example of a mosaic pixel pattern in the 7 × 7 direction. In this example, the points of 1 indicate pixels to be processed as a mosaic, and the points of 0 indicate pixels to which the original color remains without being processed.

Explanation of step 12) The shift memory is moved pixel by pixel at each pixel position for processing the mosaic image pattern, and the color information stored in the shift memory at each pixel position is assigned to the input image memory corresponding to that position. To do. The 7 × 7 mosaic image pattern shown in FIG. 9 will be taken as an example and described in more detail with reference to FIG. The color information initially stored in the shift memory is the color information at the center position (pixel position of 0) of the mosaic image pattern. Next, the shift memory is moved upward by 3 pixels so as to be at the first pixel position (pixel position of 1) of the mosaic image pattern. At this position, the color stored in the shift memory is transferred to the input image data memory image memory (1), (2), (3). As a result, the color at the center position (0 pixel) is also arranged at the 1 pixel position. Next, the shift memory is moved to two pixel positions, that is, one pixel to the left and one pixel below. And
At this position, the color stored in the shift memory is transferred to the input image data memory. As a result, the color at the center position (0 pixel) is also transferred to the 2 position. If this is repeated up to 22 in FIG. 10, it means that the color of the central pixel is arranged in all the pixels designated by the mosaic image pattern.

Explanation of step 13) 1 is added to the direction data Idirection. This makes Idir
ection becomes 1.

Description of Step 14) Returning to Step 9, when the direction data is 1, that is, the edge component is in the diagonal direction from the upper right to the lower left, the mosaic shape processing corresponding to FIG. 8 is performed.

The above processing is repeated until the direction data exceeds 4, and this processing is completed.

In the first embodiment described above, four types of simple convolution operators were considered as the edge detection operator, but the edge detection operator is not limited to this and, for example, an operator as shown in FIG. 11 is also conceivable. . These are correlated with each pixel value in the 3 × 3 area centered on (i, j) of the input image, and the operator showing the maximum correlation is found to find the edge direction near the (i, j) image position. Sex can be detected.

In this case, it is desirable to input the number corresponding to each of these operators even for the shape of the mosaic.

Furthermore, in the first embodiment described above, if the size of the mosaic shape in the first area indicating the main subject area is small and the mosaic shape in the second area indicating the background area is large, the main subject area becomes the background area. This is preferable because it will be reproduced with a finer structure compared to.

Furthermore, due to the difference between the mosaic shape pattern corresponding to the case where the maximum directionality detection value in the main subject area is less than or equal to the threshold and the one in the background area, the main subject area and the background area still have a mosaic direction. It can also be distinguished from sex, which has the effect of clarifying the subject.

Also, without specially preparing a pixel shape pattern corresponding to the case where the maximum direction detection value is equal to or less than the threshold value as in FIG. 9, one is randomly selected from the pixel shape patterns corresponding to OP1 to OP4. , You may use it.

In this case, if the maximum detection direction is less than or equal to the threshold value in step 6, the number indicating the selected pixel shape pattern is written in the image memory (i, j) instead of 0, and the I direction in step 8 is set to 1. Just do it.

In the above embodiment, the case where the area is divided into two has been described, but this may be divided into a plurality of areas and the processing may be different in each area.

As described above, according to the present embodiment, the degree of discreteness is high as compared with the conventional regular rectangular mosaic processing, and a mosaic processed image that makes use of the edge directionality of the input image is possible. Processing becomes possible and more pictorial images can be realized.

In addition, it can provide new images that were not available in the "dark room processing" that was performed by conventional silver halide systems and electronic plate making of printing systems, and also increased the degree of freedom in drawing.

Further, if the representative values of the parameters necessary for the processing are set in advance, the image processing process can be unmanned, so no specialist is required, and it is expected to be popularized in a laboratory. Although the image processing unit is used as the image processing unit in the present embodiment, it is needless to say that the same effect can be obtained by using a general-purpose computer such as a minicomputer.

In the present embodiment, the image is input from the film, but even if the image data is directly fetched from the subject from a still video camera, a video camera, etc., the magnetic floppy disk,
The same effect can be obtained by inputting image data via a recording medium such as a magnetic tape, an optical disk, or a bubble memory.

[Effect]

As described above, according to the present invention, a creative image can be expressed as compared with the conventional regular mosaic processing, and the mosaic processing that makes the best use of the characteristics of the original image information can be performed.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of the flow of a density signal when the present invention is applied to a color scanner, FIG. 2 is a block diagram showing in more detail a part of the image processing section of FIG. 1, and FIG. 4A and 4B are diagrams showing the state of the look-up table used in the description of FIG. 4, FIGS. 4A and 4B are flow charts of the embodiment, and FIG. 5 is that of FIG. FIG. 6 is a flow chart showing details of a part, FIG. 6 is a flow chart showing details of a part of FIG. 4, FIG. 7 is an example of an operator for detecting an edge direction, and FIGS. 8 and 9 are mosaic shapes. An example diagram, FIG. 10 is a diagram for explaining a mosaic process by pixel shift, and FIG. 11 is a diagram showing a second edge direction detection operator. 1 ... Input drum, 2 ... Logarithmic conversion circuit 3 ... AD converter, 4 ... Input signal processing unit 5 ... Image processing unit, 6 ... Output signal processing unit, 7 ... DA converter, 8 ... Modulator 9 …… Laser light source, 10 …… Output drum 11 …… Control processor 12 …… CPU memory 13 …… Parameter controller 14 …… Calculator, 15 …… Parameter memory 16 …… Parameter setting I / O 17 …… Image controller 18 …… Processor, 19 …… Flag map memory 20 to 27 …… Image memory 28 to 35 …… Lookup table 36 …… Image data I / O

Claims (2)

[Claims] 1. An image processing method in which an original image is divided into a plurality of divided areas and conversion processing is performed using a representative value in the divided areas, an edge direction of the image is detected, and the edge direction of the image is detected. An image processing method characterized by changing conversion processing within a division area. 2. The image processing method according to claim 1, wherein the original image is divided into a plurality of areas, and the shape of the divided area is changed according to each area.