JPH04352936A - Image processing method and system using the same - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image processing technology for shading correction of fundus images obtained with a fundus camera or remote sensing images.

0002

[Background Art] When dark areas, so-called shading, partially exist in a medical image such as a fundus image or a remote sensing image, the shading is corrected to make the image uniformly bright as a whole. It is desirable to output the For this reason, conventionally, a blurred image obtained by smoothing an input image is used as a mask image for shading correction, and shading correction is performed by obtaining a ratio, that is, a ratio image, with the input image. Alternatively, a fixed shading correction mask image corresponding to the imaging optical system is prepared, and a ratio image with the input image is obtained to perform shading. This method is disclosed in Japanese Patent Application No. 1-197645 and Japanese Patent Application No. 197645.
-199749 etc.

0003

[Problems to be Solved by the Invention] When photographing a fundus image, etc., if the lighting is not appropriate, very dark areas may appear in the image. When this image is digitized, the digital values of dark areas may end up being the smallest constant value that can be expressed. Conversely, if there is a very bright part in the image (in the case of a fundus image, a papillary part), the digital value of this part may be the maximum constant value that can be expressed.

[0004] The values of both these parts are not appropriate. Therefore, even if you apply shading correction to the image,
There is a problem that a satisfactory result cannot be obtained in a portion having this inappropriate value. For example, when dealing with a color image, the hue and saturation used to indicate the properties of the color become unrealistic values in this area.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide an image processing method capable of obtaining a high-quality shading-corrected image even for a color image having information-missing portions.

[0006]

[Means and operations for solving the problems] An image processing method of the present invention that solves the above problems includes the steps of detecting a portion where information is missing in each of RGB colors in a color image, and detecting a portion where information is missing in each color of RGB in a color image. The present invention is characterized by comprising a step of interpolating and a step of performing shading correction on the interpolated image, and performs appropriate shading correction even on images with missing information by eliminating the influence thereof.

[0007]

[Embodiment] An embodiment in which the present invention is applied to an ophthalmological image processing system will be described below. Of course, it goes without saying that the present invention can also be applied to other medical image or remote sensing image processing systems. FIG. 1 shows the overall configuration of a fundus image processing system. In the figure, reference numeral 1 denotes a fundus camera device, which can take a color fundus image using a built-in CCD camera. 2 is a slide scanner that reads the fundus photograph that has already been taken, digitizes it, and inputs it. 3 is a computer that performs various image processing, analysis and diagnosis, and also controls the entire system, and is a personal computer, workstation, or the like. 4 is a memory device such as a magneto-optical disk for storing a large amount of image data and output results; 5 is a CRT or liquid crystal color display; and 6 is a printer. 7 is an input device such as a keyboard or a mouse, through which an operator gives various instructions to the system. The processed image is displayed in various formats on the color display 5 or printed out as a hard copy by the printer 6. Doctors can make a diagnosis by looking at it.

[0008] The fundus is photographed with the CCD camera of the fundus camera 1, and the obtained fundus image is digitized with an A/D converter and stored in the memory device 4, or a previously taken fundus photograph etc. is read with the slide scanner 2 and stored in the memory. It is stored in the device 4. Next, the computer 3 performs shading correction on the color image. The outline of the processing in this embodiment is as follows. First, on the RGB space of the input color fundus image, information is interpolated by providing information on colors for which information remains for colors for which information is missing. After that, it is converted into a color space consisting of a hue image, a brightness image, and a chroma image, and the brightness image in this is subjected to grayscale value correction (shading correction).
Perform inverse conversion to a GB color image.

FIG. 2 is a flowchart showing detailed steps of image processing in this embodiment. Hereinafter, the processing procedure will be explained in detail according to the flowchart shown in FIG. First, in step 11, the original fundus color image A from the image memory 5 is read into the image processor 4. Next, in step 12, a portion where the values input for each RGB color of the original fundus color image A are the maximum or minimum value and information is missing is detected. This detected result will be referred to as detection result B. FIG. 3 shows an example in which the values of B and R have become minimum values due to strong shading in the input fundus image. FIG. 3(a) is the input fundus image, and FIG.
(b) is a diagram showing the RGB components.

[0010] In step 13, RGB of color image A is
The color information of the detection result B portion for each color is interpolated. This interpolated image will be called C. Specifically, Figure 3
As shown in (c), the information of R corresponding to the missing portions of G and B with information missing is copied while maintaining the distance between RGB. In the figure, the R signal corresponding to the missing portion is translated downward and copied.

In step 14, the color image C is converted into a color space consisting of a hue image, a brightness image, and a chroma image. This color space is called HSI, HSV, HSL, HCL, etc., but HSV will be used for explanation here. H represents hue, S represents saturation, and V represents brightness. The image after this conversion will be called D.

In step 15, grayscale value correction, that is, shading correction, is performed on V representing brightness in color image D. As shown in FIG. 5, this operation is performed only on the brightness V, and the hue H and saturation S are left unchanged. This corrected color image will be called E.

Specifically, as shown in FIG. 4, smoothing processing is performed on the brightness information image (a) to create a blurred image (b).
generate. Next, image (a) is divided by image (b) to calculate the ratio of the two, and the image (
c) generate. Then, each pixel of image (c) is multiplied by a certain value to obtain image (d) having the desired brightness. You can freely set the brightness of the entire image depending on the multiplier value. Let this corrected image be E. Note that the method of correcting the gradation value is not limited to the above method, and if the imaging optical system such as a fundus camera is known, a method of correcting it in accordance with the known imaging optical system may be used. In step 15, as shown in FIG. 5, only the brightness information V is corrected, and the hue H and saturation S information are not manipulated, so the hue H and saturation S information will not be affected. do not have. Then, in step 16, the color image E is converted back into the RGB color space to obtain a color image F. In step 17, the corrected color image F is output. This includes operations such as outputting an image to the display 5 or printer 6 and storing it in the memory device 4.

[0014] According to this embodiment, even if the image is partially missing information, it is possible to appropriately correct the shading values, and in this case, the hue and saturation information does not change, so the image is highly diagnostic. can be obtained. In addition, the brightness after correction can be freely set, so when comparing multiple images taken under different lighting (for example, images of the same patient with different shooting dates and times or with different imaging equipment), each image can be adjusted freely. It is possible to correct the brightness so that the brightness is the same. This allows doctors to make more accurate diagnoses.

Next, as another embodiment of the present invention, as shown in FIG.
An example of application to an ophthalmological system with a similar configuration is shown below. As in the previous embodiment, the fundus image is stored in the image memory 5, and then the computer 3 performs shading correction on the color image. The outline of the processing in this embodiment is as follows. First, an input fundus image is converted into a color space consisting of a hue image, a brightness image, and a chroma image. Then, correction of the grayscale values (shading correction) is performed on the brightness image, and the hue and saturation values of the areas where information is missing are estimated from the hue and saturation of the areas where no information is missing. Interpolate by giving a value (for example, an average value). After that, inverse conversion is performed to an RGB color image.

The details of the process are shown in the flowchart of FIG. In the flowchart of FIG. 6, step 21 and step 11, step 22 and step 12, step 23 and step 14, step 26 and step 15, step 27 and step 16, step 28 and step 1
7 corresponds and performs similar processing.

In this embodiment, step 1 in the previous embodiment is
Since the processing in step 3 (RGB information interpolation) is not performed, Figure 7
As shown in the figure, the portion where the information on hue H and saturation S is missing (detection result b) becomes an inappropriate value. Therefore, in this embodiment, in steps 24 and 25, as shown in FIG. 7, the information missing portions of the image of hue H and saturation S are interpolated by giving the average value of other portions as an estimated value. In addition,
The estimated values of hue and saturation given to the information-missing portions are not limited to average values, but values approximated by functions of hue and saturation may be given depending on the nature of the image. By performing such processing, this embodiment can also obtain the same effects as the previous embodiment.

[0018]

As described above, according to the present invention, even for an image in which information is partially missing, the grayscale values can be corrected in the same way as conventional shading correction, and a high-quality corrected image can be obtained.

[Brief explanation of drawings]

FIG. 1 is a system configuration diagram of a fundus image processing system according to an embodiment.

FIG. 2 is a flowchart showing the image processing procedure of the embodiment.

FIG. 3 is a diagram of an input fundus image, its RGB components, and interpolated RGB components.

FIG. 4 is an explanatory diagram of gray value correction.

FIG. 5 is an explanatory diagram of gray value correction in a color image.

FIG. 6 is a flowchart of an image processing procedure in another embodiment.

FIG. 7 is an explanatory diagram of interpolation on an HSV image.

[Explanation of symbols]

1 Fundus camera 2 Slide scanner 3 Computer 4 Memory device 5 Display 6. Printer 7 Input device

Claims (9)

[Claims] 1. A first step of detecting a portion where information is missing in each RGB color in a color image, a second step of interpolating the information of the missing portion, and a third step of performing shading correction of the interpolated image. An image processing method comprising: 2. The image processing method according to claim 1, wherein in the first step, a portion whose value is a maximum value or a minimum value of an input device in any one of RGB colors is detected as a missing portion. 3. The image processing method according to claim 1, wherein the second step provides information of a color for which information is missing in the RGB space for a color for which information remains. 4. In the second step, in the color space converted image, values estimated from the hue and saturation of the portions where no information is missing are given to the portions where information is missing regarding hue and saturation. 1. The image processing method described in 1. 5. The third step comprises: smoothing the original image to generate a blurred image; calculating a ratio between the original image and the blurred image; and multiplying the result by a constant. Item 1. Image processing method according to item 1. 6. Claims 1 to 5, wherein the image is a fundus image.
Image processing method described. 7. Input means for inputting a color image, means for detecting a portion where information is missing in each RGB color in the input color image, means for interpolating information on the missing portion, An image processing system comprising means for performing shading correction and output means for outputting an image subjected to the correction. 8. The image processing system according to claim 7, wherein said input means includes a medical image input device. 9. The image processing system according to claim 8, wherein said output means includes a display or a printer.