JP2023153733A - Image processing device - Google Patents

Abstract

To provide an image processing method which visualizes fine irregularities of a lower layer portion covered with blood.SOLUTION: An image processing method converts an original image into an RGB color space, generates a gray plane by performing filtering monochromization for each pixel of an image converted into the RGB color space, and enhances only a luminance ratio (Y1) of a region having a specific phase by using a color enhancement filter when generating the gray plane. Further, the method converts the original image into a YUV color space, YCbCr color space or YPbPr color space, compares the converted image with the gray plane at luminance (Y) for each pixel and employs the higher (brighter) one to make it a color image, and performs sharpening processing on the color image.SELECTED DRAWING: Figure 5

Description

The present invention relates to an image processing method for visualizing fine irregularities in a lower layer portion covered with blood, for example.

Depending on the purpose of use, the brightness and contrast of an image are emphasized. For example, Patent Document 1 discloses an image processing device whose purpose is to discover blood vessels or bleeding sites during endoscopic surgery, for example.
Specifically, an image acquisition unit that acquires a captured image including a subject image obtained by irradiating the subject with illumination light from a light source unit; A visibility enhancement unit that performs processing to relatively increase visibility; a detection unit that detects a blood area that is a blood area in a captured image based on color information of the captured image; A notification processing unit is disclosed that performs notification processing regarding a region.

Patent Document 2 lists brightness, saturation, red emphasis, skin color correction, contrast, and scratch noise addition as adjustment items, and an example of setting red emphasis is an image of autumn leaves, and an example of increasing brightness is an image of an aurora. Images taken are listed.

Patent Document 3 describes a vascular system imaging method that uses highlighting to highlight veins in red and arteries in blue and highlight them with different highlights in order to distinguish between pixels representing the venous vasculature and pixels representing the arterial vasculature. The method is disclosed.

Patent Document 4 discloses an image processing method for endoscopic images. This image processing method divides the image into a first area where the oxygen saturation exceeds a certain value and a second area where the oxygen saturation is below a certain value, performs color balance processing on the first area, and changes the color to the second area. It is disclosed that a gain process is performed to change the color balance process, and a red emphasis process that emphasizes a red component is performed as the color balance process.

Patent Document 5 discloses that a red-enhancing polarizing optical lens integrally provided with a polarizing filter makes it easier to identify bleeding points and finds minute bleeding points.

WO2018-235179 publication JP 2020-99014 Publication Special Publication No. 2020-508713 Japanese Patent Application Publication No. 2014-50453 WO2020-031502 publication

Patent Document 1 discloses a process of increasing the saturation and brightness of the color of blood vessel areas, but this is intended to make it easier to see the location of bleeding points and blood vessels, and as in the present application, it is difficult to see due to blood. The purpose is not to make the lower part (the lower part covered by blood) clearer, but the processing for this purpose is to widen the gradation of the red area to make the red lighter and increase the contrast. The processing was not disclosed.

Patent Documents 2 to 5 also relate to processing to make a red part stand out from other parts, and do not visualize the image of the part under the blood that is hidden by the blood (red part).
That is, the gist of the present invention is to perform processing to erase red, rather than processing to emphasize red, and none of the patent documents disclose this point.

The present invention does not make a particular color, for example, a red part, stand out from other parts, but rather removes the negative effects of red and makes the part underneath visible. To this end, the image processing method according to the present invention converts the original image into an RGB color space, performs filtering and monochrome conversion for each pixel of the image converted into the RGB color space, and generates a gray plane. In this case, a color emphasis filter is used to increase only the brightness rate (Y1) of an area having a specific hue. Furthermore, the original image is converted to the YUV color space, YCbCr color space, or YPbPr color space, and the converted image and the gray plane are compared for each pixel in terms of brightness (Y), and the higher (brighter) one is adopted to determine the color. image, and perform sharpening processing on this color image.

When the purpose is to visualize minute irregularities in the lower layer covered by blood, a red emphasis filter is used as the color emphasis filter, and only the brightness rate (Y1) of the area with a red hue close to that of blood is visualized. enhance

Furthermore, it is conceivable that color enhancement of a specific color using a color enhancement filter can be made closer to white by increasing the brightness of the specific color.

According to the image processing method of the present invention, for example, the lower part covered with blood can be visualized during surgery, so the surgery can be performed while checking the location of small irregularities and wounds, increasing safety. Can be done.

Original image with bleeding Black and white image of the original image with bleeding A black and white image with increased brightness in the red area of the original image with bleeding. A color image that combines the replaced luminance signal (Y1), the difference between the luminance signal and the blue component (U), and the difference between the luminance signal and the red component (V) Image obtained by sharpening the composite color image

Embodiments of the present invention will be described below based on the accompanying drawings.
Figure 1 is an original image with bleeding, and as can be seen from this image, it is not possible to see what the lower part covered with blood looks like.

FIG. 2 shows a black and white image obtained by converting the original image into an RGB color space and converting it into a monochrome image through filtering, in which red color has high saturation but low brightness and is dark. In other words, it can be seen that the reason why the lower part covered with blood cannot be visualized is because the brightness is low.

Therefore, as shown in FIG. 3, a black-and-white image (brightness rate: Y1) is created in which the red hue area (bleeding area) of the original image is brightened using a red emphasis filter.

In the filtering monochrome processing, weights are determined for each hue of R (red), Ye (yellow), G (green), C (cyan), B (blue), and M (magenta) from an RGB image and converted to gray scale. do.
Assuming that the weights for each hue are Rf (red), Yef (yellow), Gf (green), Cf (cyan), Bf (blue), and Mf (magenta), in this example, Rf = 100%, Yef = 60 %, Gf = 40%, Cf = 60%, Bf = 20%, and Mf = 80%. Incidentally, the human luminance sensitivity to red expressed by the hue wheel is Rf = 40%.
It is also possible to weaken a specific color by specifying a negative ratio for f.

The following calculation is performed, assuming that Y is the black and white value (brightness rate) of a pixel having R, G, and B values.
Here, max=the maximum value of RGB, mid=a value that is neither the maximum nor the minimum value of RGB, and min=the minimum value of RGB. However, if there is an equivalent value, mid will be the same as max or min.
(1) If max==mid==min, use that value as Y and end the calculation. (2) Start from Y=0. (3) If max==R, Y+=Rf・(max-mid)
(4) When max==B Y+=Yef・(max-mid)
(5) When max==G Y+=Gf・(max-mid)
(6) When min==R Y+=Cf・(max-mid)
(7) When min==B Y+=Bf・(max-mid)
(8) When min==G Y+=Mf・(max-mid)
(9) Two of (3) to (8) are always applicable (three if there are equivalent values)
(10) Y+=min
(11) If Y>255, Y=255
(12) If Y<0 then Y=0

For example, the brightness rate of red expressed on the hue wheel is 40%, but this brightness rate is multiplied by 2.5 to make it 100%. By converting the brightness rate of red to 100%, the red area in which the red gradation in the original image was configured from black to red is converted to a wide gradation from black to red to white. By converting the red region into a wide gradation, it can be seen that minute contrasts can be visualized even in the same black and white photograph compared to FIG. 2.

In the present invention, the magnification of the brightness factor of the red color described above is an adjustable parameter, and is not limited to 2.5 times, but may be 2.0 times, etc.
When there are many blue components such as veins, it is possible to adjust the difference in hue of the blood image by varying the magnification of not only R (red) but also M (magenta). , Ye (yellow), G (green), C (cyan), B (blue), and M (magenta).By setting the magnification according to the hue, it is possible to transmit areas with any hue. It is.

Next, the brightness rate (Y1) of the red region of the monochrome image brightened by the red emphasis filter is replaced with the brightness (Y) of the red region of the YUV color space, YCbCr color space, or YPbPr color space of the original image. That is, the brightness rate (Y) of the red area of the original image is multiplied by 2.5. FIG. 4 shows a color image created by combining the replaced luminance factor (Y) with the UV, CbCr, or PbPr of the original image.

As shown in FIG. 4, it can be seen that by increasing the brightness of the red region, parts that were dark in the original image become brighter and the gradation becomes wider, and the unevenness of the part under the blood becomes visible. However, at the same time, red appears to be a fluorescent color at the border of the red area.

The reason for the above is that when an image is decomposed into YUV and only the red area is brightened (for example, 2.5 times) using a red emphasis filter for Y (luminance), the area that was originally colored from black to bright red changes from black to red. It has a wide gradation range that includes pink with white in it, giving it a fluorescent color image mixed with bright pink, which makes it easy to overlook small irregularities and scratches.

In order to solve the above problem, the widened gradation is returned to the natural brightness range. Specifically, PN.6867563, PN. By using 6893068, CLAHE, LISr (registered trademark), etc., sharpening processing is performed to uniformize the contrast adaptively.

FIG. 5 shows a color image after the sharpening process, and when compared with the original image, it can be seen that the unevenness of the portion covered with blood has been visualized.

In the embodiment, a description has been given of visualization of a portion covered with blood during surgery by increasing the brightness rate of the red area and widening the gradation of the red area, but the brightness rate of areas other than red may be increased. In particular, it can be used to sharpen the lower part of an image area where the saturation is high and the brightness is low.

Claims (3)

An image processing method consisting of the following steps.
(1) Convert the original image to RGB color space.
(2) Perform filtering and monochrome conversion for each pixel of the image converted to the RGB color space to generate a gray plane.
(3) When generating the gray plane, a color emphasis filter is used to increase only the brightness rate (Y1) of an area having a specific hue.
(4) Convert the original image to YUV color space, YCbCr color space, or YPbPr color space.
(5) Compare the image converted to the YUV color space, YCbCr color space, or YPbPr color space and the gray plane in terms of brightness (Y) for each pixel, and select the higher (brighter) one.
(6) Perform sharpening processing on the color image obtained in (5). 2. The image processing method according to claim 1, wherein the color emphasis filter is a red emphasis filter, and increases only the brightness rate (Y1) of a region having a red hue close to blood. 3. The image processing method according to claim 2, wherein the red emphasis by the red emphasis filter makes the brightness of red the same as that of white.

Images