JPH07121703A - Image processing method - Google Patents

Abstract

PURPOSE:To restore an original image with high accuracy by finding a degradation function. CONSTITUTION:The output of an image input device 1 is connected to a computer 2, and an input device 3, image display device 4 and image preservation device 5 are connected to the computer 2. Inside the computer 2, a small area including one edge part at least is designated in an image inputted from the input device 1, a certain function such as a regular function containing a parameter is assumed as the degradation function, the image is restored while changing parameters T and (n), and degree of image restoration is calculated, and the entire image is restored by using the degradation functions of parameters Tmax and Nmax which improve most the degree of image restoration.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a field of digital image processing and relates to an image processing method for estimating and restoring a deterioration function of a deteriorated image.

[0002]

2. Description of the Related Art When applying a Wiener filter, a general inverse filter, a restricted least squares filter, etc., which are various methods of image restoration, it is first necessary to determine a deterioration function.
The deterioration function is most ideally obtained by analytically obtaining from the physical phenomenon that causes the deterioration, or when the measuring device is at hand, by directly measuring the input-output relationship and estimating the deterioration characteristic.

However, when these methods cannot be used, the deterioration function must be estimated directly from the deteriorated image. The following method is known for estimating this deterioration function.

(A) Estimation of point spread function (b) Estimation from line spread function (c) Estimation from edge spread function

Particularly in the case of (c), if there is a sharp edge portion in the original image, a method of differentiating the edge portion to obtain a line spread function and determining a deterioration function using an image reconstruction method is available. Are known.

[0006]

However, when obtaining the deterioration function from the edge portion in the original image, it is necessary to extract the edge portion from the original image, but since the original image is deteriorated, the edge portion is removed. Difficult to extract.

Further, even if the edge can be extracted, the edge portion is not always a straight line, so that it is difficult and difficult to obtain the deterioration function using the image reconstruction method.

The object of the present invention is to solve the above-mentioned problems,
An object of the present invention is to provide an image processing method capable of easily obtaining a deterioration function and restoring an original image with high accuracy.

[0009]

According to an image processing method of the present invention for achieving the above object, a small area including at least an edge portion is designated in a deteriorated image, and a parameter is set in the small area. The image restoration is performed while changing the degradation function including the parameter, the image restoration degree corresponding to the parameter is obtained, and the degradation function including the parameter is selected based on the image restoration degree, and the entire image subjected to the degradation is selected. It is characterized by performing the image restoration of.

[0010]

According to the image processing method having the above-described structure, a small area including at least an edge portion is specified in the deteriorated image, and the image restoration is performed in the small area by using the deterioration function of a certain parameter. The image restoration degree is obtained, and the image restoration and the operation for obtaining the image restoration degree are repeated while changing the parameters, and the image restoration of the entire image that has been deteriorated is performed by using the deterioration function of the parameter with the best image restoration degree. To do.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail based on the illustrated embodiments. FIG. 1 is a block circuit configuration diagram of an embodiment. The output of an image input device 1 such as a slide scanner is connected to a computer 2, and the computer 2 has an input device 3, an image display device 4 such as a display, a magnetic disk and the like. The image storage device 5 is connected.

The image read into the computer 2 from the image input device 1 is instructed by the input device 3 as shown in FIG.
The process shown in the flowchart of FIG. The processed image is stored in the image storage device 5 as needed.

Referring to the flow chart of FIG. 2, first, at step 11, from the input image as shown in FIG. 3, at least at the edge portion R where the deterioration can be clearly identified, or at the place where the operator most desires to restore the image. A region R including one edge is designated by a rectangular ROI or the like, and step 1
In step 2, the area designated in step 11 is cut out as shown in FIG. Next, assuming a Gaussian window function ω _G (x) based on a normal distribution as the following equation as the deterioration function, in step 13, as shown in FIG. 5, the Gaussian window function ω _G (x) of the Gaussian window function ω _G (x) The Wiener filter is applied by using the parameter T indicating the diameter of n and the parameter n indicating the shape of the Gaussian window. Here, ω _R (x) indicates a rectangular window whose width is the parameter T. ω _G (x) = ω _R (x) ・ exp (-2n ² · ‖x‖ ² / T ² ) ・ ・ ・ (1)

Next, in step 14, binarization is performed by the discriminant analysis method as shown in FIG. 6 to calculate the degree of separation. The binarization by this discriminant analysis method is performed as follows.

(A) First, a histogram of image density values is created. (b) The predetermined threshold value is set to k, and the pixel is divided into two groups of pixels whose density value is equal to or larger than the threshold value k and pixels whose density value is smaller than the threshold value k, and are classified into classes 1 and 2, respectively. the number of pixels omega ₁ (c), each class 1,2, ω _2, the average density value of M _1, M _{2, 1} a variance sigma, and sigma _2, also the average density value of all pixels as M _T, Intraclass variance σ _W ² and interclass variance σ _B ² are calculated. Within-class variance σ _W ²
And the interclass variance σ _B ² are given by the following equation, and the degree of separation is defined by the ratio σ _B ² / σ _W ² .

Σ _W ² = ω ₁ σ ₁ ² + ω ₂ σ ₂ ² (2) σ _B ² = ω ₁ · (M ₁ −M _T ) ² + ω ₂ · (M ₂ −M _T ) ² = _{ω 1 · ω 2 · (M} 1 -M 2) 2 ··· (3)

For example, when the original image having a sharp edge portion as shown in FIG. 7 is deteriorated and becomes an image with a smooth edge portion as shown in FIG. 8, the images of FIG. 7 and FIG. When a density histogram of an image is created, graphs as shown in FIGS. 9 and 10 are obtained. In the histogram of the original image, the two classes 1 and 2 are clearly separated, the interclass variance σ _B ² is large, and the separation σ _B ² / σ _W ² is also large. On the other hand, in the histogram of the deteriorated image, the interclass variance σ _B ² is small and the separation σ _B ² / σ _W ² is also small. Therefore,
The degree of separation σ _B ² / σ _W ² is a measure showing the degree of restoration of an image having an edge.

(D) (b) and (c) are repeated for all thresholds k, and the threshold k at which the separation σ _B ² / σ _W ² is maximum and the separation σ _B ² / σ _W ² at that time are Desired.

Further, in step 15, the operations of steps 13 to 15 are repeated while changing the parameters T and N of the Gaussian window function ω _G (x) within a predetermined range. As a method of moving the parameters T and N, first, a rough search is performed in which the parameters are moved with a large width to obtain an approximate number of the parameters T and N, and the width is sequentially narrowed to perform a fine search, thereby reducing the calculation time. Can be planned. Further, in step 16, parameters T _max and N _max that maximize the degree of separation σ _B ² / σ _W ² are obtained from the result obtained in step 15.

Here, when noise is emphasized when the Wiener filter is applied in step 13, the threshold value k can be satisfactorily obtained even if evaluation is performed using only the separation degree σ _B ² / σ _W ^2. There are things you can't do. In this case, the binarized image in step 14 is labeled with the connected components,
Among those having the smallest number of labels, the one having the largest degree of separation σ _B ² / σ _W ² may be adopted.

Finally, in step 17, as shown in FIG. 11, the parameters T _max and N obtained in step 16 are calculated.
_The Gaussian window function ω _G (x) including _max is used to restore the entire image using a conventional image restoration method such as a Wiener filter.

In this embodiment, instead of directly obtaining the deterioration function, a certain function such as, for example, a normal function is assumed as the deterioration function and the parameter is changed to determine the deterioration function, so that the deterioration degree with the best restoration is obtained. The function can be determined, and highly effective image restoration can be performed.

[0023]

As described above, in the image processing method according to the present invention, it is not necessary to extract the edge portion from the original image, and the deterioration function can be obtained only by designating the area including the edge portion. The original image can be accurately restored. Moreover, it is possible to restore an image that does not require a high-level technique such as a reconstruction method.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment.

FIG. 2 is a flowchart showing processing means.

FIG. 3 is an explanatory diagram of processing means.

FIG. 4 is an explanatory diagram of processing means.

FIG. 5 is an explanatory diagram of processing means.

FIG. 6 is an explanatory diagram of processing means.

FIG. 7 is a graph showing the image density at the edge of the original image.

FIG. 8 is a graph showing the image density at the edge of the deteriorated image.

FIG. 9 is a graph showing a histogram of image density at an edge of an original image.

FIG. 10 is a graph showing a histogram of image density at the edge of a deteriorated image.

FIG. 11 is an explanatory diagram of processing means.

[Explanation of symbols]

 1 image input device 2 computer 3 input device 4 image display device 5 image storage device

Claims (2)

[Claims] 1. A small area including at least an edge portion in a degraded image is designated, image restoration is performed while changing a deterioration function including a parameter in the small area, and image restoration corresponding to the parameter is performed. The image processing method is characterized in that the deterioration function is obtained, the deterioration function including the parameter is selected based on the image restoration degree, and the entire image subjected to the degradation is restored. 2. The image processing method according to claim 1, wherein the image restoration degree is obtained using a class separation degree of a discriminant analysis method.