JPH08107512A - Method for adjusting sharpness of image - Google Patents

Abstract

PURPOSE: To adjust sharpness with a large convolution by comparatively small circuit scale by smoothing a reduced image, magnifying the image to an original size, weighting the difference between the magnified image and an original image and adding the weighted difference to the source image. CONSTITUTION: This method is provided with image memories IM1-IM3 and a memory INM for input, and one or two systems of the outputs of these memories are selected by a multiplexer MUX and inputted to a computing element CAL. The output of the computing element CAL is selectively inputted to either one of memories IM1-IM3. Then, a camera C for input is connected to the memory INM for input and a fetched image is temporarily held in the memory INM. Then, the original image is reduced by prescribed reduction factor, the reduced image is smoothed, the smoothed image is magnified by magnification factor corresponding to the reduction factor, and the difference between the magnified image and the original image is added to the original image with prescribed weight. Thus, since large sharpening is executed, high-speed processing can be provided by the small scale circuit.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image sharpness adjusting method, and more particularly to a sharpening method of various sharpness in a large size convolution, that is, a sharpness adjusting method.

[0002]

2. Description of the Related Art Sharpening is a method of edge enhancement by restoring high-frequency components, and by taking a large convolution, more global processing can be realized. However, the hardware corresponding to a large convolution size has a large circuit scale and becomes an expensive system. Also, image processing by software cannot obtain a practical processing speed.

[0003]

SUMMARY OF THE INVENTION The present invention was devised to solve the above-mentioned problems of the prior art, and it is possible to realize sharpness adjustment of a large convolution with a relatively small circuit scale. The purpose is to provide a method of adjustment.

[0004]

An image sharpness adjusting method according to the present invention smoothes a reduced image, enlarges it to its original size, and weights the difference between the enlarged image and the original image. The result of multiplying by is added to the original image, and as a result,
Large-scale sharpening is executed. Then, the sharpness is continuously adjusted by changing the weight.

[0005]

According to the image sharpness adjusting method of the present invention, the reduced image is smoothed and enlarged to the original size, and the difference between the enlarged image and the original image is weighted. Can be further added to the original image.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an image sharpness adjusting method according to the present invention will now be described with reference to the drawings. In FIG.
In an example of an image processing apparatus for implementing the method,
The image memories IM1 to IM3 and the input memory INM are provided, and one or two systems of outputs of these memories are selected by the multiplexer MUX and input to the arithmetic unit CAL. The output of the arithmetic unit CAL is selectively input to any of the memories IM1 to IM3 by the selector SEL. On the other hand, an input camera C is connected to the input memory INM,
The captured image is temporarily stored in the memory INM.

FIG. 2 is a flow chart showing the procedure for adjusting the sharpness, the details of which will be described below. [Step 2-1] First, the original image is taken from the camera C, held in the memory INM, and then transferred to the memory IM1. At this time, the arithmetic unit CAL does not execute the arithmetic processing and passes the data as it is.

[Step 2-2] The original image in the memory IM1 is read out pixel by pixel, and is written into the memory IM2 while thinning out to generate a reduced image. FIG. 3 shows this thinning-out situation. Pixels are extracted at regular intervals, for example, every 8 pixels (marked with a cross in the figure), and other pixels are ignored. In the image processing device, this thinning can be performed by performing writing to the memory IM2 at regular intervals. The arithmetic unit CAL does not perform an arithmetic operation during the reduction processing, and passes data.

[Step 2-3] The reduced image in the memory IM2 is read pixel by pixel, and the processing result is written in the memory IM3 while being smoothed by the arithmetic unit CAL. The smoothing convolution size is typical, such as 3x3, and can be implemented in small circuits. The smoothing performed here is the density average of the pixels other than the central pixel, and in the convolution of FIG. 4 (specifying each pixel by the symbols A to I), E = (A + B + C + D + F + G + H + I) / 8 is calculated.

[Step 2-4] The smoothed image in the memory IM3 is read pixel by pixel, and the processing result is written in the memory IM2 while being further smoothed by the arithmetic unit CAL. This smoothing is the same as the processing performed in step 2-3.

[Step 2-5] The smoothed image in the memory IM2 is read out pixel by pixel, enlarged and written in the memory IM3, and restored to the size of the original image. The enlargement processing is realized by writing the read pixel a plurality of times and further writing the same line a plurality of times. Writing to the same pixel is realized by repeatedly giving the same address when reading the memory IM2.

[Step 2-6] Weighted addition is performed on the image in the memory IM3 and the original image stored in the memory IM1, and the result is written in the memory IM2. The pixel density of the original image is Do, the pixel density of the IM3 image is D, the weight is α,
If the density of the processing result is Dr, the weighted addition is expressed by the following equation. Dr = (1 + α) Do−αD Here, when α = 1, Dr = Do + (Do−D), and the value of D is approximately equal to the smoothed image of 32 × 32 convolution, and Dr is 32 as a result. The image is sharpened by x32 convolution. Further, as α is decreased, the second term on the right side of Dr = Do + α (Do-D) is decreased, the sharpness is decreased, and the image gradually approaches the original image.

The smoothing for the reduced image described above has the same effect as the smoothing for the large convolution, and the large sharpening is realized by adding the large smoothed image and the original image. And as hardware, 3x3
The purpose is achieved with a normal scale such as convolution. Furthermore, the sharpness can be adjusted by adjusting the coefficient α,
In digital copiers, facsimiles, digital photographs, etc., the sharpness can be arbitrarily set according to the application or the user's preference.

[0014]

As described above, the image sharpness adjusting method according to the present invention smoothes a reduced image, enlarges it to its original size, and weights the difference between the enlarged image and the original image. Since a large-scale sharpening is executed by adding the product multiplied by to the original image, it has an excellent effect that high-speed processing can be realized by a small-scale circuit.

[Brief description of drawings]

FIG. 1 is a block diagram showing an image processing apparatus used for implementing a method of the present invention.

FIG. 2 is a flowchart showing an embodiment of the method of the present invention.

FIG. 3 is a conceptual diagram showing a thinning state in the method of the present invention.

FIG. 4 is a conceptual diagram showing 3 × 3 convolution.

[Explanation of symbols]

 MUX multiplexer IM1 image memory 1 IM2 image memory 2 IM3 image memory 3 INM memory CAL calculator SEL selector C camera

Claims (4)

[Claims] 1. An original image is reduced at a predetermined reduction rate, smoothing is performed on the reduced image, the smoothed image is enlarged at an enlargement rate corresponding to the reduction rate, and an enlarged image is obtained. A method for adjusting the sharpness of an image, which comprises adding a difference from the original image to the original image with a predetermined weight. 2. The image sharpness adjusting method according to claim 1, wherein the reduction ratio is 1/8 and the enlargement ratio is 8. 3. The method for adjusting the sharpness of an image according to claim 1, wherein the sharpness of the image is increased by setting the weight to 1. 4. The image sharpness adjusting method according to claim 1, wherein the image is smoothed with a weight of 1/2 or less.