Vascular enhancement methods, endoscopic equipment, storage media, and computer program products

By processing the combined brightness and color information of endoscopic images, and combining filtering and enhancement techniques, the problem of insufficient extraction of fine blood vessels in endoscopic images was solved, achieving better blood vessel enhancement effects.

CN123071218APending Publication Date: 2026-11-10SONOSCAPE MEDICAL CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510590582.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-07
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing endoscopic image processing methods are not very effective in enhancing blood vessels, especially in the extraction and enhancement of fine blood vessels, and cannot effectively utilize the vascular features acquired by the endoscopic system.

Method used

We use the combined brightness and color information of endoscopic images to extract and enhance vascular texture features. By combining brightness and color filtering kernels, and texture enhancement and color enhancement methods, we reconstruct enhanced vascular images.

Benefits of technology

It improves the completeness of fine blood vessel texture feature extraction, significantly enhances the blood vessel texture features in endoscopic images, and improves the visual effect of the images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123071218A_ABST
    Figure CN123071218A_ABST
Patent Text Reader

Abstract

This invention provides a method for enhancing blood vessels, an endoscopic device, a storage medium, and a computer program product, comprising: acquiring an endoscopic image; decomposing the endoscopic image to obtain a target brightness image under a target brightness channel and a target color image under a target color channel; extracting vascular texture features from the endoscopic image based on the target brightness image and the target color image to obtain an initial texture image, and enhancing the initial texture image to obtain an enhanced texture image; and reconstructing the image based on the target image to obtain a vascular enhanced image, wherein the target image includes the enhanced texture image. This invention can extract and enhance vascular texture features in an image based on the joint information of the image's brightness and color, so that the obtained vascular texture and edge response have a positive effect on the visual effect of the final acquired image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of endoscopy technology, and more specifically to a method for enhancing blood vessels, an endoscopic device, a storage medium, and a computer program product. Background Technology

[0002] An endoscope system is a medical electro-optical instrument integrating optical, mechanical, and electronic technologies. It includes an illumination device, an imaging device, an image processing device, and a display device. The endoscopic imaging process typically involves: first, generating illumination light using the illumination device, which is then guided through an optical fiber within the endoscope to illuminate the internal cavity. The illumination light is reflected by the tissue within the cavity to the objective lens at the tip of the endoscope, where it is converted into a digital image signal by a photoelectric sensor. This image signal is then transmitted via cable to an image processor for further processing, and finally, the processed image is output to the display.

[0003] In endoscopic image processing, image enhancement of blood vessels is a crucial step. Current methods for blood vessel enhancement mainly include three approaches: The first method uses digital filtering to extract texture and edge features from the image, performs a global nonlinear transformation on it, and then superimposes it onto the original image to obtain the final enhanced blood vessel image. The second method distinguishes between vascular and non-vascular regions based on color in the image, and then performs stretching transformations on both to improve the contrast between blood vessels and surrounding tissues. The third method decomposes the image into brightness and color components, and then performs edge texture enhancement and color contrast enhancement on each component separately. However, these three approaches only consider blood vessel enhancement from the perspective of digital image processing, and the enhancement effect still needs improvement. Summary of the Invention

[0004] The present invention was proposed in view of the above-mentioned problems. The present invention provides a method for enhancing blood vessels, an endoscopic device, a storage medium, and a computer program product.

[0005] According to one aspect of the present invention, a method for enhancing blood vessels is provided, comprising: acquiring an endoscopic image; decomposing the endoscopic image to obtain a target brightness image under a target brightness channel and a target color image under a target color channel; extracting vascular texture features from the endoscopic image based on the target brightness image and the target color image to obtain an initial texture image, and performing texture enhancement on the initial texture image to obtain an enhanced texture image; and performing image reconstruction based on the target image to obtain a vascular enhanced image, wherein the target image includes the enhanced texture image.

[0006] Optionally, extracting vascular texture features from the endoscopic image based on the target brightness image and the target color image includes: performing joint filtering on the target brightness image and the target color image based on a brightness filtering kernel and a color filtering kernel to extract vascular texture features from the endoscopic image.

[0007] Optionally, the target color channel includes a first color channel and a second color channel, and the target color image includes a first color image under the first color channel and a second color image under the second color channel; the joint filtering operation is performed using the following formula:

[0008]

[0009] Among them, I t (x,y) represents the texture intensity of the pixel at coordinates (x,y) in the initial texture image, O represents the neighborhood of the pixel at coordinates (x,y), and I Y(center) I represents the brightness value of the center pixel of neighborhood O in the target brightness image. c1(center) I represents the color value of the center pixel of neighborhood O in the first color image. c2(center) I represents the color value of the center pixel of neighborhood O in the second color image. Y (i,j) represents the brightness value of the pixel at coordinates (i,j) in the target brightness image. c1 (i,j) represents the color value of the pixel at coordinates (i,j) in the first color image. c2 (i,j) represents the color value of the pixel at coordinates (i,j) in the second color image; G1 represents the brightness filter kernel, G c This represents the color filter kernel, where δ1 is the standard deviation of the luminance filter kernel, and δ c W represents the standard deviation of the color filter kernel. xy This represents the normalization coefficient of the filter kernel.

[0010] Optionally, texture enhancement is performed on the initial texture image to obtain an enhanced texture image, including: performing a global nonlinear transformation on the initial texture image using the following formula to achieve texture enhancement:

[0011]

[0012] In the formula, I represents the texture intensity of the pixel with coordinates (x, y) in the enhanced texture image. t(x,y) represents the texture intensity of the pixel at coordinates (x,y) in the initial texture image, a represents the first preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the first scale edge and the lower limit of the preset texture intensity range corresponding to the second scale edge, b represents the second preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the second scale edge and the lower limit of the preset texture intensity range corresponding to the third scale edge, c represents the maximum texture intensity in the initial texture image, f1 represents the preset mapping value corresponding to the first preset texture intensity a, f2 represents the preset mapping value corresponding to the second preset texture intensity b, and f3 represents the preset mapping value corresponding to the maximum texture intensity c; wherein, the edge type of blood vessels is divided into first scale edge, second scale edge and third scale edge according to the preset size range, and the upper or lower limit of the size range corresponding to the first scale edge, second scale edge and third scale edge increases sequentially.

[0013] Optionally, before performing image reconstruction based on the target image to obtain a blood vessel enhanced image, the method further includes: performing color enhancement on the target color image to obtain an enhanced color image; wherein the target image also includes the enhanced color image.

[0014] Optionally, color enhancement is performed on the target color image to obtain an enhanced color image, including: inputting the color value of each pixel in the target color image into a preset blood vessel mask model to obtain a blood vessel mask image; performing contrast stretching on the target color image based on the blood vessel mask image to obtain an enhanced color image; wherein, the blood vessel mask model is used to determine whether the pixel corresponding to the input color value belongs to a blood vessel.

[0015] Optionally, the preset blood vessel mask model is constructed by: acquiring a first sample endoscopic image; extracting multiple blood vessel color sample values ​​from the first sample endoscopic image, wherein the blood vessel color sample values ​​are the color values ​​of the pixels belonging to the blood vessels in the target color channel; and performing Gaussian fitting on the multiple blood vessel color sample values ​​to obtain the preset blood vessel mask model.

[0016] Optionally, the target color channel includes a first color channel and a second color channel, and the preset blood vessel mask model is represented by the following formula:

[0017]

[0018] In the formula, Indicates the first standard color value. P(I) represents the second standard color value. c1, I c2 ) indicates that the color value in the first color channel is I. c1And the color value in the second color channel is I c2 The similarity between pixels and blood vessels, δ m p0 is the standard deviation of the preset blood vessel mask model, p0 is the preset minimum similarity threshold, and I m This indicates that the color value in the first color channel is I. c1 And the color value in the second color channel is I c2 The mask values ​​of the pixels; where the first standard color value Second standard color value and standard deviation δ m Determined through Gaussian fitting.

[0019] Optionally, before performing image reconstruction based on the target image to obtain a blood vessel enhanced image, the method further includes: extracting structural edge features from the endoscopic image based on the target brightness image to obtain an initial structural edge image; performing structural edge enhancement on the initial structural edge image to obtain an enhanced structural edge image; wherein the target image also includes the enhanced structural edge image.

[0020] Optionally, the endoscopic image is decomposed to obtain a target brightness image in the target brightness channel and a target color image in the target color channel, including: determining the illumination spectrum of the illumination light emitted by the endoscopic system when acquiring the endoscopic image; and decomposing the endoscopic image based on the illumination spectrum to obtain a target brightness image in the target brightness channel and a target color image in the target color channel.

[0021] Optionally, the endoscopic image is an image in a first color space, which includes at least two color channels. The endoscopic image is decomposed based on the illumination spectrum to obtain a target brightness image in a target brightness channel and a target color image in a target color channel. This includes: assigning a corresponding target channel weight to each of the at least two color channels based on the illumination spectrum, wherein the target channel weight of the color channel corresponding to the principal component light in the illumination spectrum is greater than the target channel weights of other color channels; and converting the endoscopic image from the first color space to a second color space based on the target channel weights corresponding to each of the at least two color channels to obtain the target brightness image in the target brightness channel and the target color image in the target color channel; wherein the second color space includes the target brightness channel and the target color channel.

[0022] Optionally, before assigning corresponding target channel weights to each of the at least two color channels based on the illumination spectrum, the method further includes: for the illumination spectrum, performing the following weight determination operation to determine the target channel weight reassembly corresponding to the illumination spectrum: presetting at least two sets of initial channel weights based on the illumination spectrum, wherein each set of initial channel weights includes initial channel weights that correspond one-to-one with at least two color channels, and the initial channel weights included in any two sets of initial channel weights are not completely identical; acquiring a second sample endoscopic image; decomposing the second sample endoscopic image according to the at least two sets of initial channel weights to obtain at least two initial brightness images under the target brightness channel, wherein the at least two initial brightness images correspond one-to-one with at least two sets of initial channel weights; selecting one set of initial channel weights from the at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum according to the texture effect evaluation parameters of each initial brightness image, wherein the target channel weight reassembly includes the target channel weights corresponding to each of the at least two color channels.

[0023] Optionally, the texture effect evaluation parameters include texture response intensity. Based on the texture effect evaluation parameters of each initial brightness image, a set of initial channel weights is selected from at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum, including: obtaining the texture response intensity corresponding to each of the at least two initial brightness images; and using the set of initial channel weights corresponding to the initial brightness image with the strongest texture response intensity as the target channel weight reassembly corresponding to the illumination spectrum.

[0024] Optionally, obtaining the texture response intensity corresponding to each of the at least two initial brightness images includes: for each of the at least two initial brightness images, obtaining at least one region of interest in the initial brightness image; calculating the local edge contrast of each region of interest in the initial brightness image; and calculating the average value of the local edge contrast of the at least one region of interest to obtain the texture response intensity corresponding to the initial brightness image.

[0025] Optionally, the first color space is RGB space; when the wavelength corresponding to the principal component light is located on the short-wavelength side: W b >W g >W r When the wavelength corresponding to the principal component light is on the longer wavelength side: W b <W g <W r Among them, W b +W g +W r =1, W b W represents the target channel weight corresponding to channel B. g W represents the target channel weight corresponding to channel G. rThe target channel weight is the value corresponding to channel R.

[0026] According to another aspect of the present invention, an endoscopic device is also provided, comprising: a processor and a memory, wherein the memory stores computer program instructions, which are executed by the processor to perform the above-described method for enhancing blood vessels.

[0027] According to another aspect of the present invention, a storage medium is also provided, on which program instructions are stored, which, when executed, are used to perform the above-described vascular enhancement method.

[0028] According to another aspect of the present invention, a computer program product is also provided, including computer program instructions that, when executed, are used to perform the vascular enhancement method as described above.

[0029] Because endoscopic images have low sharpness, the brightness difference between fine blood vessels and surrounding tissues at the edge junction is small, while the color difference is large. Therefore, this embodiment of the invention extracts and enhances vascular texture features in the image based on the combined information of brightness and color of the endoscopic image. While making full use of brightness information to ensure the extraction of high-contrast vascular texture features, it can also combine color information to extract low-contrast fine blood vessel texture features, which is beneficial to improving the completeness of vascular texture feature extraction. This allows the fine blood vessel texture features in the endoscopic image to be significantly enhanced, thereby giving the final vascular enhanced image (i.e., the endoscopic image after vascular enhancement processing) a better visual effect. Attached Figure Description

[0030] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0031] Figure 1 A schematic flowchart of a vascular enhancement method according to an embodiment of the present invention is shown;

[0032] Figure 2 A schematic diagram of a vascular texture enhancement curve according to an embodiment of the present invention is shown;

[0033] Figure 3 A schematic diagram of statistical results of blood vessel color distribution according to an embodiment of the present invention is shown;

[0034] Figure 4 A schematic diagram of a blood vessel color enhancement curve according to an embodiment of the present invention is shown;

[0035] Figure 5 A schematic block diagram of an endoscopic device according to an embodiment of the present invention is shown. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0037] During the imaging process of an endoscopic system, the image quality output to the display is highly correlated with the various components of the system. First, the illumination spectrum (which includes information such as the wavelength components and proportions of each wavelength) determines the endoscopic system's ability to extract features from intraluminal tissues and the image contrast of those tissues. Second, due to the small size of the endoscope tip, the area of ​​the photoelectric sensor placed at the tip must be designed to be sufficiently small. This prevents endoscopic systems from improving image quality by increasing sensor pixel size or designing telephoto lens modules, as is done with traditional SLR cameras. Therefore, endoscopic systems must compensate for hardware limitations through image processing algorithms. Furthermore, besides common white light imaging, there are various special light imaging techniques, such as narrowband imaging. All of these imaging techniques require deep adaptation of the endoscopic system's image processing algorithms.

[0038] In related technologies, vascular enhancement is only considered from the perspective of digital image processing, and it cannot be specifically enhanced based on the vascular features acquired during imaging by the endoscopic system. Therefore, the vascular enhancement effect needs further improvement. Specifically, the brightness difference between microvessels and mucosal background tissue at the edge is very small, so it is difficult to extract the vascular texture with weak contrast using only brightness information. This invention provides a vascular enhancement method that can extract and enhance vascular texture features in endoscopic images acquired by an endoscopic system based on the combined information of brightness and color of the endoscopic image. This method can extract microvascular texture and improve the completeness of vascular texture extraction.

[0039] Please see Figure 1 The diagram shown is a schematic flowchart of a vascular enhancement method according to an embodiment of the present invention. The vascular enhancement method of the present invention may include, but is not limited to, steps S110, S120, S130, and S140.

[0040] Step S110: Obtain endoscopic images.

[0041] For example, an endoscope system can be used to acquire images of cavities in a biological body (such as the digestive tract, bronchi, bladder, etc.). Typically, endoscopic images may also include blood vessels, which can be categorized into deep, intermediate, and superficial vessels based on their depth within the tissue cavity. Superficial vessels are usually fine and difficult to identify, but their distribution and morphology are crucial for endoscopic diagnosis. Therefore, extracting and enhancing these fine vessels from endoscopic images is particularly important.

[0042] Step S120: Decompose the endoscopic image to obtain the target brightness image under the target brightness channel and the target color image under the target color channel.

[0043] For example, the target brightness image represents the brightness component of the endoscope image, and the target color image represents the color component of the endoscope image. That is, the target brightness image represents the brightness information of the endoscope image, and the target color image represents the color information of the endoscope image. The target color channel may include a first color channel and a second color channel; correspondingly, the target color image may include a first color image under the first color channel and a second color image under the second color channel, with the first color image and the second color image respectively representing the first color information and the second color information of the endoscope image.

[0044] Generally, endoscopic images are in RGB format. Therefore, in this embodiment of the invention, the color space of the endoscopic image acquired by the endoscopic system can be RGB space. The target brightness image and target color image corresponding to the endoscopic image can be obtained by converting the endoscopic image in RGB space to YUV or YCbCr space. Taking the conversion to YUV space as an example, in YUV space, the target brightness image can be the Y channel component (hereinafter referred to as "Y component"), and the target color image can include the U channel component (hereinafter referred to as "U component") and the V channel component (hereinafter referred to as "V component"). Those skilled in the art will understand that endoscopic images can be converted from RGB space to YUV space based on the conversion formulas corresponding to the digital television standard ITU-R BT.601 or ITU-R BT.709. Generally, the Y component can be quantized to the range of [16, 235], and the U and V components can be quantized to the range of [16, 240]. This embodiment of the invention does not specifically limit the conversion formula for converting endoscopic images from RGB space to YUV space. The conversion formula can be directly applied using the aforementioned conversion formulas provided by the digital television standards, or an optimized conversion formula can be obtained by adjusting the parameters based on the aforementioned conversion formulas of the digital television standards according to the actual needs of the vascular enhancement image to be acquired, or it can be defined manually. In another specific embodiment, when converting the endoscopic image to YCbCr space, the target luminance image can be the Y component, and the target color image can include the Cb and Cr components. Of course, in yet another specific embodiment, as will be described below, the endoscopic image can also be converted to a custom color space, as long as the separation of the luminance and color signals of the endoscopic image can be achieved.

[0045] Step S130: Extract vascular texture features from the endoscopic image based on the target brightness image and the target color image to obtain an initial texture image, and perform texture enhancement on the initial texture image to obtain an enhanced texture image.

[0046] For example, extracting vascular texture features from endoscopic images can be achieved by filtering the target brightness image and target color image using filters such as bilateral filtering, locally weighted least squares (LWLS) filtering, and guided image filtering (Guided Filter), which will not be elaborated further. This embodiment of the invention does not limit the specific texture enhancement method; methods for enhancing the initial texture image can include, for example, the Laplacian operator, high-pass filtering, adaptive enhancement, etc. It is understood that since the target brightness image can represent the brightness information of the endoscopic image, and the target color image can represent the color information of the endoscopic image, extracting vascular texture features from the target brightness image and target color image is equivalent to jointly extracting vascular texture features from the brightness and color information of the endoscopic image.

[0047] Step S140: Perform image reconstruction based on the target image to obtain the enhanced blood vessel image corresponding to the endoscopic image, wherein the target image includes the enhanced texture image.

[0048] It is understandable that the enhanced vascular image is the image obtained after performing vascular enhancement processing on the endoscopic image.

[0049] For example, the target image may include an enhanced texture image in YUV space, and may also include enhanced / unenhanced target color images corresponding to the target color channels, enhanced / unenhanced structural edge images corresponding to the target luminance channels, and target luminance images corresponding to the target luminance channels, etc., in YUV space. After obtaining the target image in YUV space, the target image can be converted from YUV space to RGB space to obtain a blood vessel enhancement image in RGB space. Those skilled in the art will understand that the target image can be converted from YUV space to RGB space based on the conversion formulas corresponding to the digital television standard ITU-R BT.601 or ITU-R BT.709, and the blood vessel enhancement image in RGB space can be obtained after image reconstruction of the color space-converted target image.

[0050] Because endoscopic images have low sharpness, the brightness difference between fine blood vessels and surrounding tissues at the edge junction is small, while the color difference is large. Therefore, this embodiment of the invention extracts and enhances vascular texture features in the image based on the combined information of brightness and color of the endoscopic image. While making full use of brightness information to ensure the extraction of high-contrast vascular texture features, it can also combine color information to extract low-contrast fine blood vessel texture features, which is beneficial to improving the completeness of vascular texture feature extraction. This allows the fine blood vessel texture features in the endoscopic image to be significantly enhanced, thereby giving the final vascular enhanced image (i.e., the endoscopic image after vascular enhancement processing) a better visual effect.

[0051] Optionally, vascular texture features in the endoscopic image are extracted based on the target brightness image and the target color image, including:

[0052] The target brightness image and target color image are jointly filtered based on the brightness filter kernel and the color filter kernel to extract the vascular texture features in the endoscopic image.

[0053] For example, a joint filtering technique (such as joint bilateral filtering) can be used to process the brightness information of the target brightness image and the color information of the target color image to extract vascular texture features. The brightness filtering kernel can mainly process the brightness information of the target brightness image, which can determine the grayscale features of the endoscopic image. The color filtering kernel can process the color information of the target color image, which can determine the color features (such as hue, saturation, etc.) of the endoscopic image. In a specific embodiment, a luminance domain kernel radius and a color domain kernel radius can be defined. The luminance domain kernel radius defines the size of the luminance domain neighborhood and is used to calculate the luminance weight of the luminance difference between pixels. The color domain kernel radius defines the size of the color domain neighborhood and is used to calculate the color weight of the color difference between pixels. The luminance weight of the target pixel in the target luminance image and the color weight of the target pixel in the target color image are calculated based on the luminance domain kernel radius and the color domain kernel radius, respectively. The product of the color weight and the luminance weight is defined as the joint weight. For any target pixel, its filtered output value can be calculated by accumulating the luminance values ​​(or color values) of all pixels in the neighborhood, multiplying them by the corresponding joint weight, and then dividing by the sum of all weights. The above steps are repeated for each pixel, and the final filtered result can be used as the initial texture image. The features of this initial texture image include vascular texture features.

[0054] By employing a combined brightness-color filter to extract vascular texture from an image, the brightness and color information can be combined to extract vascular texture from the brightness information, which is beneficial for extracting the texture features of low-contrast microvessels and improving the completeness of vascular texture feature extraction.

[0055] Optionally, the target color channel includes a first color channel and a second color channel, and the target color image includes a first color image under the first color channel and a second color image under the second color channel;

[0056] The joint filtering operation is performed using the following formula (1):

[0057]

[0058] Formula (2) is used to calculate the normalization coefficient W of the filter kernel. xy I t (x,y) represents the texture intensity of the pixel at coordinates (x,y) in the initial texture image, O represents the neighborhood of the pixel at coordinates (x,y) in the target brightness image, and I Y(center) I represents the brightness value of the center pixel of neighborhood O in the target brightness image. c1(center) I represents the color value of the center pixel of neighborhood O in the first color image. c2(center) I represents the color value of the center pixel of neighborhood O in the second color image. Y(i,j) I represents the brightness value of the pixel at coordinates (i,j) in the target brightness image. c1(i,j) I represents the color value of the pixel at coordinates (i,j) in the first color image. c2(i,j) G1 represents the color value of the pixel at coordinates (i,j) in the second color image; G1 represents the brightness filtering kernel. c The color filter kernel is represented by δ1, and the standard deviation of the luminance filter kernel is δ. c W is the standard deviation of the color filter kernel. xy This represents the normalization coefficient of the filter kernel.

[0059] For example, the luminance filter kernel can be processed after the color space of the endoscopic image is converted to a luminance-chrominance space. The converted color space can be, for example, YUV space or YCbCr space. Taking YUV space as an example again, the target color channel in YUV space can include a U channel (first color channel) and a V channel (second color channel). More specifically, the U component under the U channel can be recorded as the first color image, the V component under the V channel can be recorded as the second color image, and the Y component under the Y channel can be recorded as the target luminance image. Each pixel of the initial texture image corresponds one-to-one with each pixel of the target luminance image, each pixel of the first color image, and each pixel of the second color image.

[0060] Let the coordinates of any pixel in the initial texture image be (x, y), and the coordinates of a pixel within its neighborhood O be (i, j). Then, in the target brightness image, the first color image, and the second color image, there exists a pixel corresponding to the pixel at coordinates (i, j) in the initial texture image. Furthermore, in the target brightness image, the first color image, and the second color image, the pixel corresponding to the pixel at coordinates (i, j) in the initial texture image also has coordinates (i, j) in its respective image. Before jointly filtering the target brightness image and the target color image using a brightness filter kernel and a color filter kernel, the neighborhood radii of each pixel in the target brightness image and the neighborhood radii of each pixel in the target color image can be preset according to the quality requirements of the blood vessel enhancement image to be acquired. The neighborhood radii of each pixel in the target brightness image and the neighborhood radii of each pixel in the target color image can be equal, and the neighborhood radii of each pixel in the first color image and the second color image belonging to the target color image are equal.

[0061] By using both brightness and color filtering kernels to extract vascular texture features, the extracted vascular texture feature information can be made as complete as possible.

[0062] Optionally, texture enhancement is performed on the initial texture image to obtain an enhanced texture image, including:

[0063] Texture enhancement is achieved by performing a global nonlinear transformation on the initial texture image using the following formula (3):

[0064]

[0065] In the formula, I represents the texture intensity of the pixel with coordinates (x, y) in the enhanced texture image. t (x,y) represents the texture intensity of the pixel with coordinates (x,y) in the initial texture image, a represents the first preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the first scale edge and the lower limit of the preset texture intensity range corresponding to the second scale edge, b represents the second preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the second scale edge and the lower limit of the preset texture intensity range corresponding to the third scale edge, c represents the maximum texture intensity in the initial texture image, f1 represents the preset mapping value corresponding to the first preset texture intensity a, f2 represents the preset mapping value corresponding to the second preset texture intensity b, and f3 represents the preset mapping value corresponding to the maximum texture intensity c.

[0066] The edge types of blood vessels are divided into first-scale edges, second-scale edges and third-scale edges according to a preset size range, with the upper or lower limit of the size range corresponding to the first-scale edge, second-scale edge and third-scale edge increasing sequentially.

[0067] For example, the edge type of blood vessels can be divided according to a preset size range. Specifically, the sizes of different types of blood vessels can vary, while blood vessels of the same type can fall within a certain preset size range. For example, the diameter of capillaries can range from 5 micrometers to 10 micrometers, the diameter of arterioles and venules can range from tens of micrometers to hundreds of micrometers, the diameter of medium-sized arteries and veins can range from 1 millimeter to several millimeters, the diameter of large arteries can range from 2.5 centimeters to 3 centimeters, and the diameter of large veins can be around several centimeters. The aforementioned size ranges of blood vessels of each type can be used as the basis for dividing the preset size range. The size ranges of the preset first-scale edge, second-scale edge, and third-scale edge can be adaptively set according to the types of blood vessels contained in the image acquisition area corresponding to the endoscopic image. Alternatively, they can be set in combination with the types of blood vessels to be retained in the image acquisition area and the texture effect requirements of each type of blood vessel. For example, blood vessels at the first-scale edge can be defined as image noise, that is, blood vessels at the first-scale edge may not be retained in the blood vessel enhancement image to be acquired. This is not specifically limited here.

[0068] Please see Figure 2 As shown, it is a schematic diagram of a vascular texture enhancement curve according to an embodiment of the present invention. Figure 2 In the provided embodiments, the horizontal axis represents the texture response intensity, and the vertical axis represents the preset mapping value corresponding to each texture response intensity. The texture response intensity and its corresponding preset mapping value can be set according to the texture effect requirements of the blood vessel enhancement image to be acquired. This embodiment of the invention does not specifically limit the mapping relationship between the initial texture response intensity and the enhanced texture response intensity. Figure 2The function corresponding to the vascular texture enhancement curve can be referred to in the above formula (3). The first preset texture intensity 'a' serves as a common segmentation point between the first-scale edge and the second-scale edge. It can be used to divide the first-scale edge and the second-scale edge according to the texture intensity to define the texture intensity range of the first-scale edge and the second-scale edge. The first preset texture intensity 'a' can be used as the upper limit of the preset texture intensity range corresponding to the first-scale edge, and also as the lower limit of the preset texture intensity range corresponding to the second-scale edge. Similarly, the second preset texture intensity 'b' serves as a common segmentation point between the second-scale edge and the third-scale edge. It can be used to divide the second-scale edge and the third-scale edge according to the texture intensity to define the texture intensity range of the second-scale edge and the third-scale edge. The second preset texture intensity 'b' can be used as the upper limit of the preset texture intensity range corresponding to the second-scale edge, and also as the lower limit of the preset texture intensity range corresponding to the third-scale edge. The maximum texture intensity 'c' in the initial texture image can be directly obtained from the initial texture image, which will not be elaborated here. In this embodiment of the invention, the texture intensity of pixels corresponding to blood vessels at the second-scale edge is subjected to a linear transformation with a slope of 1, while the texture intensity of pixels corresponding to blood vessels at the first-scale and third-scale edges is subjected to a linear transformation with a slope of less than 1. In another embodiment, the texture intensity of pixels corresponding to blood vessels at the first-scale and third-scale edges may be subjected to a linear transformation with a slope of less than 1, while the texture intensity of pixels corresponding to blood vessels at the second-scale edge may be subjected to a linear transformation with a slope greater than 1.

[0069] Although the grayscale changes of vascular texture are dramatic, the amplitude of the changes is relatively small, and it has the characteristic of local dynamic range concentration. By suppressing the texture of blood vessels at small and large scale edges to a certain extent, and enhancing the texture of blood vessels at medium scale edges, it is possible to effectively avoid excessive amplification of image noise and help prevent artifacts from appearing at edges with strong texture response intensity.

[0070] Optionally, before performing image reconstruction based on the target image to obtain a blood vessel enhancement image, the method further includes:

[0071] Perform color enhancement on the target color image to obtain the enhanced color image;

[0072] The target image also includes an enhanced color image.

[0073] For example, taking a target color image in YUV space as an example, color saturation can be enhanced by increasing the absolute values ​​of the color values ​​corresponding to the U component (first color image) and V component (second color image). This can also be achieved by adjusting the weights of the U and V components; for example, increasing the weight of the U component can make the acquired enhanced blood vessel image more yellow. When performing image reconstruction to obtain an enhanced blood vessel image, the color-enhanced target color image can also participate in image reconstruction as the target image.

[0074] When the color of blood vessels is similar to the color of the surrounding tissue, color enhancement can improve the contrast between vascular and non-vascular areas in the image, thereby effectively improving the visual effect of the final enhanced vascular image.

[0075] Optionally, color enhancement is performed on the target color image to obtain an enhanced color image, including:

[0076] Input the color values ​​of each pixel in the target color image into a preset blood vessel mask model to obtain a blood vessel mask image;

[0077] The target color image is contrast-stretched based on the blood vessel mask image to obtain an enhanced color image;

[0078] Among them, the blood vessel mask model is used to determine whether the pixel corresponding to the input color value belongs to a blood vessel.

[0079] For example, taking a target color image in YUV space as an example, the Y component (first color image) or the U component (second color image) can be input separately into a preset blood vessel mask model. Alternatively, the color features of the first color image corresponding to the U channel and the second color image corresponding to the V channel can be fused, and the target color image can be input as a whole into the preset blood vessel mask model. Specifically, the image type input into the preset blood vessel mask model is determined based on the type of the training image of the blood vessel mask model. The color values ​​of each pixel in the target color image are input into the blood vessel mask model. The blood vessel mask model can determine whether each input pixel belongs to a blood vessel based on the comparison between the color values ​​of each input pixel and the statistical / fitting results of the color values ​​of the pixels corresponding to blood vessels in YUV space. The output result obtained by the blood vessel mask model is the blood vessel mask image corresponding to the endoscopic image. The blood vessel mask image can be used to distinguish between blood vessel regions and non-blood vessel regions in the target color image. Based on this distinction result, the contrast of the target color image can be stretched.

[0080] By contrast stretching the vascular and non-vascular regions of each color image in the color space, the vascular and non-vascular regions can be distinguished as much as possible, thereby enhancing the vascular region.

[0081] Optionally, the preset blood vessel mask model is constructed in the following way:

[0082] Acquire the first sample endoscopic image;

[0083] Multiple blood vessel color sample values ​​are extracted from the first sample endoscopic image. The blood vessel color sample value is the color value of the pixel belonging to the blood vessel in the target color channel.

[0084] Gaussian fitting is performed on multiple blood vessel color sample values ​​to obtain a preset blood vessel mask model.

[0085] For example, the preset blood vessel mask model can be obtained based on a large number of first sample endoscopic images. Similar to the endoscopic images in the aforementioned embodiments, the first sample endoscopic images can also be images acquired by an endoscopic system targeting a biological cavity. The first sample endoscopic image can contain multiple blood vessels, and the color value of the pixels belonging to the blood vessels in the target color channel is the blood vessel color sample value. The target color channel can be the U channel and / or V channel in the YUV space. For the acquired first sample endoscopic images, statistical methods can be used to statistically analyze the color values ​​of the pixels corresponding to the blood vessels in the first sample endoscopic images in the U channel and / or V channel in the YUV space. Based on the statistical analysis results, the color value range of the pixels corresponding to the blood vessels in the U channel and / or V channel is obtained to construct the blood vessel mask model. Alternatively, a data fitting method, such as a Gaussian fitting method, can be used to fit the color values ​​of the pixels corresponding to the blood vessels in the U channel and / or V channel in the YUV space to construct the blood vessel mask model. It should be noted that multiple blood vessel color sample values ​​can be obtained based on a single first sample endoscopic image. Therefore, there is no limit to the number of first sample endoscopic images, as long as a sufficient number of blood vessel color sample values ​​can be extracted.

[0086] Please see Figure 3 As shown, it is a schematic diagram of the statistical results of blood vessel color distribution according to an embodiment of the present invention. Figure 3 In the diagram, the horizontal axis represents the color value of the pixel in the first color image (U component) of the blood vessel in YUV space, and the vertical axis represents the color value of the pixel in the second color image (V component) of the blood vessel in YUV space. Specifically, Figure 3 It is based on the Gaussian model of blood vessels in a two-dimensional color space obtained by extracting 100,000 blood vessel color sample values ​​from several first sample endoscopic images and performing Gaussian fitting. This Gaussian model can be used as a preset blood vessel mask model.

[0087] By pre-modeling blood vessel colors using a large number of blood vessel sample values, the distribution characteristics of blood vessels in the color space can be accurately obtained, thereby improving the accuracy of distinguishing blood vessels from surrounding tissues.

[0088] Optionally, the target color channel includes a first color channel and a second color channel, and the preset blood vessel mask model is represented by the following formulas (4) and (5):

[0089]

[0090]

[0091] In the formula, Indicates the first standard color value. P(I) represents the second standard color value. c1, I c2 ) indicates that the color value in the first color channel is I. c1 And the color value in the second color channel is I c2 The similarity between pixels and blood vessels, δ m p0 is the standard deviation of the preset blood vessel mask model, p0 is the preset minimum similarity threshold, and I m This indicates that the color value in the first color channel is I. c1 And the color value in the second color channel is I. c2 The mask values ​​of the pixels; where the first standard color value Second standard color value and standard deviation δ m Determined through Gaussian fitting.

[0092] For example, taking the target color channels corresponding to the target color image and the first sample endoscope image as both being U channel and V channel, the U channel is denoted as the first color channel, and the V channel as the second color channel. For the above formulas (4) and (5), the first standard color value can be the Gaussian fitting result of the color value of the pixel of the blood vessel under the first color channel (U channel), and similarly, the second standard color value can be the Gaussian fitting result of the color value of the pixel of the blood vessel under the second color channel (V channel). For any pixel A of the endoscope image, during the decomposition of the endoscope image, there is a specific pixel B in the target color image under the target color channel, which corresponds to pixel A of the endoscope image. The specific corresponding content can include spatial location information in the image (e.g., pixel coordinates), color value information (e.g., the color value of each pixel in the endoscope image after decomposition under each target color channel), etc. The preset minimum similarity threshold can be defined by the image quality requirements of the blood vessel enhancement image. In this embodiment of the invention, the preset minimum similarity threshold p0 = 0.7 is preferred. Those skilled in the art will understand that the standard deviation of the preset blood vessel mask model can also be determined by Gaussian fitting. This standard deviation can measure the dispersion of the color values ​​corresponding to the U and / or V channels of each blood vessel pixel in the YUV space. Its value is related to the number of blood vessel pixels and the color values ​​corresponding to the U and / or V channels of each pixel in the YUV space. The specific calculation process will not be elaborated here.

[0093] By acquiring a blood vessel mask image, it is possible to classify each pixel in the target color image in the color space, including blood vessels and non-blood vessels. This is beneficial for subsequent color value enhancement operations on each pixel based on its category.

[0094] Alternatively, the target color image can be subjected to contrast stretching using the following formula (6):

[0095]

[0096] In the formula, when c = c1, the above formula (6) represents nonlinear stretching of the first color image; when c = c2, the above formula (6) represents nonlinear stretching of the second color image. This represents the color value of the pixel at coordinates (x, y) in the enhanced color image, g represents the preset control parameter for stretching contrast, e is the base of the natural logarithm, and I... c (x,y) represents the color value of the pixel at coordinates (x,y) in the target color image, U represents the average pixel value of the pixel corresponding to the blood vessel in the first color image and the pixel value of the pixel corresponding to the blood vessel in the second color image, and I m(x,y)=1 indicates that the pixel at coordinates (x,y) in the target color image has a similarity to a blood vessel greater than the preset minimum similarity threshold, and therefore belongs to the blood vessel pixel category. m (x,y)=0 indicates that the pixel with coordinates (x,y) in the target color image has a similarity to the blood vessel that is less than or equal to the preset minimum similarity threshold, and belongs to the non-blood vessel pixel. The pixel with coordinates (x,y) in the target color image corresponds to the pixel with coordinates (x,y) in the enhanced color image.

[0097] For example, still taking the YUV color space of the target color image as an example, the target color image may include a first color image (U component) and a second color image (V component), wherein the target color channel corresponding to the first color image is the first color channel (U channel), and the target color channel corresponding to the second color image is the second color channel (V channel). Each pixel of the first color image and each pixel of the second color image are nonlinearly stretched using the above formula (6). When the similarity between a pixel in the target color image and a blood vessel is greater than a preset minimum similarity threshold, the expression in the first row of formula (6) is used for contrast stretching. When the similarity between a pixel in the target color image and a blood vessel is less than or equal to the preset minimum similarity threshold, no contrast stretching is performed. Wherein, in the above formula (6), I... m The specific values ​​of (x, y) can be determined using the aforementioned formula (5). For example, for any pixel C in the first color image, the color value of pixel C in the first color channel could be, for example, I. c1 Then, in the second color image, there exists a pixel D corresponding to pixel C, and the color value of pixel D in the second color channel could be, for example, I. c2 Then the mask values ​​of pixels C and D can be obtained by using the color value I of pixel C in the first color channel. c1 and the color value I of pixel D in the second color channel c2 To be determined jointly.

[0098] Please see Figure 4 As shown, it is a schematic diagram of a blood vessel color enhancement curve according to an embodiment of the present invention. Figure 4In the diagram, the horizontal axis represents the color value of the pixel in the first color image (U component) of the blood vessel in YUV space, and the vertical axis represents the color value of the pixel in the second color image (V component) of the blood vessel in YUV space. After obtaining the blood vessel mask image, the blood vessel region and non-blood vessel region in the first color image and the second color image are nonlinearly stretched in the color space. The formula for the blood vessel stretching curve can be referred to the above formula (6). Formula (6) can be understood as follows: for the non-blood vessel region, the pixel in this region retains the original color value; for the blood vessel region, the pixel in this region is stretched using the blood vessel enhancement curve. Since the mucosa color and the blood vessel color are somewhat similar but different, in this blood vessel enhancement curve, stretching is performed when the pixel color is close to the blood vessel color to improve the blood vessel saturation; compression is performed when the pixel color is significantly different from the blood vessel color to achieve mucosa fading. Thus, the color contrast between the blood vessel and the mucosa can be better improved.

[0099] Optionally, before performing image reconstruction based on the target image to obtain a blood vessel enhancement image, the method further includes:

[0100] Structural edge features are extracted from the endoscope image based on the target brightness image to obtain an initial structural edge image;

[0101] Structural edge enhancement is performed on the initial structural edge image to obtain the enhanced structural edge image;

[0102] The target image also includes an enhanced structural edge image.

[0103] For example, the color space of the target brightness image can be YUV space or YCrCb space. Taking the target brightness image (Y component / luminance component) in YUV space as an example, the pixel gradient of the target brightness image can be calculated based on edge detection algorithms, such as Sobel, Prewitt, and Canny, and the structural edge features can be extracted to obtain an initial structural edge image. For the initial structural edge image, global contrast can be enhanced by nonlinear transformations, such as gamma correction or histogram equalization, to enhance the structural edge features in the endoscopic image; local contrast enhancement can also be performed based on local image characteristics, such as local histogram edge reduction or adaptive contrast stretching of the initial structural edge image; sharpening filters, such as Laplacian filters, can also be used to sharpen the initial structural edge image. When performing image reconstruction to obtain an enhanced blood vessel image, the enhanced structural edge image can also be used as a target image in image reconstruction.

[0104] It is understandable that structural edge features affect the overall clarity of endoscopic images. Furthermore, thick blood vessels in endoscopic images are usually found in structural edge features. Therefore, this embodiment simultaneously extracts and enhances the structural edge features of the endoscope, which is beneficial to improving the overall visual effect of the enhanced blood vessel image.

[0105] Alternatively, the operation of extracting structural edge features can be performed using the following formula (7):

[0106]

[0107] In the formula, I s (x,y) represents the pixel value of the pixel at coordinates (x,y) in the initial structural edge image, Q represents the preset neighborhood of the pixel at coordinates (x,y) in the target brightness image, and r s δ represents the radius of the preset neighborhood of the pixel with coordinates (x, y) in the target brightness image. s The standard deviation of the preset Gaussian filter kernel is denoted as .

[0108] For example, structural edge features are distributed in the target brightness image, and Gaussian filtering can be applied to the target brightness image to extract structural edge features. The standard deviation of the preset Gaussian filter kernel can be chosen to distinguish the texture features of blood vessels from the structural edge features, thus avoiding spatial overlap and the introduction of noise. In the above formula (7), the radius of the preset neighborhood Q can be chosen freely, without specific limitations here.

[0109] Alternatively, the structural edge reinforcement operation can be performed using the following formula (8):

[0110]

[0111] In the formula, α is a preset structural edge enhancement factor, and I s (x,y) represents the pixel value of the pixel with coordinates (x,y) in the enhanced structural edge image.

[0112] For example, the value of the preset structural edge enhancement factor can be customized, and can be set and adjusted according to the sharpness of the initial structural edge image.

[0113] Optionally, the endoscopic image is decomposed to obtain a target brightness image in the target brightness channel and a target color image in the target color channel, including:

[0114] Determine the illumination spectrum of the illumination light emitted by the endoscope system when acquiring endoscopic images;

[0115] The endoscopic image is decomposed based on the illumination spectrum to obtain the target brightness image in the target brightness channel and the target color image in the target color channel.

[0116] For example, an endoscope system may include a light source, such as a 4-5 channel LED light source. The light source can be used to emit illumination light when acquiring endoscopic images. The illumination light may have a corresponding spectral distribution (i.e., an illumination spectrum). More specifically, the illumination spectrum can be considered as a function of the wavelengths of the components of light in the light source. The components of light in the illumination spectrum can mainly include combinations of several types of light, such as visible light and invisible light. Visible light can be, for example, red light, green light, blue light, amber light, etc., and invisible light can be, for example, ultraviolet light, near-infrared light, etc. Based on the light power ratio or luminous flux of the components of light with different wavelengths in the illumination spectrum, corresponding target channel weights can be assigned to each color channel of the endoscopic image to decompose the endoscopic image based on the target channel weights. For example, when the endoscopic image is an RGB image, if the light power ratio or luminous flux of the shorter wavelength component light is larger, a higher target channel weight can be assigned to the B channel of the endoscopic image; if the light power ratio or luminous flux of the longer wavelength component light is larger, a higher target channel weight can be assigned to the R channel of the endoscopic image.

[0117] It is understandable that light of different wavelengths has different penetration depths and scattering characteristics to tissues. Therefore, for the same object, the texture and color in images acquired under different illumination spectra typically exhibit different characteristics. Thus, the aforementioned technical solution, by acquiring the illumination spectrum, can determine the depth of the target object (i.e., the object of focus in this examination, such as blood vessels) and the dominant color distribution after imaging. Consequently, during image decomposition, more emphasis is placed on the channel values ​​containing this dominant color distribution, resulting in richer texture and detail features related to the target object in the brightness information. In other words, endoscopic images obtained under the illumination spectrum of an endoscope system can usually highlight the subject information required for the examination. Therefore, the aforementioned technical solution, by decomposing endoscopic images based on the illumination spectrum of the endoscope system, helps to allocate appropriate target channel weights to each color channel, thereby better preserving and highlighting the subject information of the endoscopic image.

[0118] Optionally, the endoscopic image is an image in a first color space, which includes at least two color channels. The endoscopic image is decomposed based on the illumination spectrum to obtain a target brightness image in the target brightness channel and a target color image in the target color channel, including:

[0119] Based on the illumination spectrum, a corresponding target channel weight is assigned to each of at least two color channels, wherein the target channel weight of the color channel corresponding to the principal component light in the illumination spectrum is greater than the target channel weight of other color channels.

[0120] Based on the target channel weight corresponding to each of the at least two color channels, the endoscopic image is converted from the first color space to the second color space to obtain the target brightness image under the target brightness channel and the target color image under the target color channel.

[0121] The second color space includes the target luminance channel and the target color channel.

[0122] For example, the first color space can be the RGB space, and at least two color channels in the first color space can include an R channel, a G channel, and a B channel. During the decomposition of the endoscopic image, the R channel, G channel, and B channel of the endoscopic image can be converted into a luminance channel (i.e., the target luminance channel) corresponding to luminance information and a color channel (i.e., the target color channel) corresponding to color information. The resulting decomposition can be a target luminance image containing luminance information under the target luminance channel and a target color image containing color information under the target color channel. The number of target color images can be the same as the number of target color channels, and the number of target luminance images can be the same as the number of target luminance channels. Typically, the number of target luminance channels is 1. Before decomposing the endoscopic image, the target channel weights of at least two color channels of the endoscopic image can be obtained. For example, the target channel weights can be an array, which includes the target channel weights of the R channel, G channel, and B channel of the endoscopic image, and the sum of the target channel weights in the array can be 1. The target channel weights can be determined based on the illumination spectrum of the illumination light when acquiring the endoscopic image. In a specific embodiment, the component light with the highest light power ratio or the highest luminous flux in the illumination spectrum can be taken as the main spectrum (i.e., the principal component light), and other component light can be taken as the non-main spectrum (i.e., the non-principal component light). If the wavelength of the main spectrum in the obtained spectral analysis results is relatively short, then the information in the B channel of the endoscopic image is the main information extracted from the main spectrum, while the information in the R channel is the background information extracted from the non-main spectrum. Based on this spectral analysis result, before decomposing the endoscopic image, a higher target channel weight can be assigned to the B channel of the endoscopic image to retain the main information as much as possible, while the background color information in the R channel is appropriately compressed (e.g., reducing the color contrast in the R channel). The conversion method from the first color space to the second color space can refer to the conversion method from the RGB color space to the YUV / YCbCr color space. It should be noted that during the process of converting the image from the RGB color space to the YUV / YCbCr color space, the channel weights of each color channel of the image are a set of fixed values. In this embodiment of the invention, the target channel weights assigned to each color channel of the endoscope image based on the illumination spectrum may not be the same as the channel weights used to convert the endoscope image from the RGB color space to the YUV / YCbCr color space. Therefore, the second color space may not be exactly the same as the YUV / YCbCr color space.

[0123] In a specific embodiment, the color space of the endoscope image is RGB space. The wavelength of the principal component light in the illumination spectrum when the endoscope image is acquired is 625-740nm. Therefore, the channel corresponding to the principal component light can be determined as the R channel, and the channels corresponding to the non-principal component light can be determined as the G channel and the B channel. The initial channel weight of the R channel can be defined to be greater than the initial channel weight of the G channel and the initial channel weight of the B channel. For example, for a set of initial channel weights, the initial channel weight corresponding to the R channel is 0.6, the initial channel weight corresponding to the G channel is 0.3, and the initial channel weight corresponding to the B channel is 0.1.

[0124] Different wavelengths of light have different penetration depths and scattering characteristics to tissues within endoscopic cavities; the longer the wavelength, the greater the penetration depth. Furthermore, different tissues within endoscopic cavities (e.g., epidermis, mucous membranes, capillaries, venules, veins, etc.) exhibit different absorption, reflection, or scattering characteristics to different wavelengths of light. Accordingly, endoscopic images of the same tissue within an endoscopic cavity acquired under illumination of different wavelengths of light will visually present different morphologies. To better highlight the tissue being observed, light with wavelengths that best highlight the tissue is typically used as the principal component light to illuminate the cavity. The information extracted from the principal component light in the illumination spectrum constitutes the main information of the endoscopic image, while the information extracted from non-principal component light (especially light with wavelengths significantly different from the principal component light) constitutes the background information. It is understood that assigning a target channel weight greater to the color channel corresponding to the principal component light than to the color channel corresponding to the non-principal component light emphasizes the main information, thereby preserving the image features of the tissue being observed as completely as possible.

[0125] Optionally, before assigning a corresponding target channel weight to each of the at least two color channels based on the illumination spectrum, the method further includes:

[0126] For the illumination spectrum, the following weighting operation is performed to determine the target channel weighting reassembly corresponding to the illumination spectrum:

[0127] At least two sets of initial channel weights are preset based on the illumination spectrum. Each set of initial channel weights includes initial channel weights that correspond one-to-one with at least two color channels. The initial channel weights included in any two sets of initial channel weights are not exactly the same.

[0128] Acquire the second sample endoscopic image;

[0129] The second sample endoscopic image is decomposed according to at least two sets of initial channel weights to obtain at least two initial brightness images under the target brightness channel, wherein the at least two initial brightness images correspond one-to-one with at least two sets of initial channel weights;

[0130] Based on the texture effect evaluation parameters of each initial brightness image, a set of initial channel weights is selected from at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum, wherein the target channel weight reassembly includes the target channel weights corresponding to each of at least two color channels.

[0131] For example, the endoscope system can emit corresponding illumination light under different acquisition scenarios. Different illumination lights can have different illumination spectra. For each illumination spectrum, at least two sets of initial channel weights can be preset for each component light of that illumination spectrum. Each set of initial channel weights includes the initial channel weights of each color channel of the endoscope image. The sum of the initial channel weights of each color channel can be 1. The initial channel weights corresponding to each color channel in each set of initial channel weights can be defined by the illumination spectrum. More specifically, the initial channel weights corresponding to each color channel in each set of initial channel weights can be determined based on the dominant spectrum in the corresponding illumination spectrum. For example, when the dominant spectrum is red light, the preset channel weight requirement can be that the initial channel weight of the color channel corresponding to red light is greater than the initial channel weight of the color channels corresponding to other component light. Then, the preset channel weight requirement can be that the values ​​of each color channel in each set of initial channel weights satisfy that the initial channel weight of the color channel corresponding to red light is greater than the initial channel weight of the color channels corresponding to other component light. For any two or more sets of initial channel weights, the initial channel weights of each color channel in each set of initial channel weights can not be completely identical.

[0132] For example, the acquisition method of the second sample endoscope image is similar to that of the first sample endoscope image, and will not be repeated here. The color space of the second sample endoscope image can be the same as that of the endoscope image, for example, it can be RGB space. Decomposing the same second sample endoscope image based on each set of initial channel weights can yield an initial brightness image under the target brightness channel. Each initial brightness image can correspond to a texture effect evaluation parameter, which can be the texture response intensity of the initial brightness image or a manually defined texture effect level. This embodiment of the invention does not impose specific limitations here. Specifically, a set of initial channel weights can be selected as the target channel weight reassembly corresponding to the illumination spectrum based on the comparison results of the texture effect evaluation parameters. Alternatively, a set of initial channel weights can be selected as the weight base, and the initial channel weights can be adjusted according to the distribution of the texture response intensity of each initial brightness image. The adjusted set of initial channel weights is then used as the target channel weight reassembly corresponding to the illumination spectrum. This embodiment of the invention does not limit the specific number of initial channel weight sets; the number of initial channel weight sets only needs to be greater than or equal to 2. The finally determined target channel weight reassembly can include the target channel weights corresponding to each color channel.

[0133] The above technical solution can decompose the image into brightness and color information based on the imaging characteristics of different wavelengths of light in the illumination spectrum of the endoscope system and the proportion of each component light, which is beneficial for accurately and completely extracting the texture and color of the target object in the endoscope image.

[0134] Optionally, the texture effect evaluation parameters include texture response intensity. Based on the texture effect evaluation parameters of each initial brightness image, a set of initial channel weights is selected from at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum, including:

[0135] Obtain the texture response intensity corresponding to at least two initial brightness images;

[0136] The initial channel weights corresponding to the initial brightness image with the strongest texture response intensity are used as the target channel weights corresponding to the illumination spectrum.

[0137] For example, the texture effect evaluation parameter can be the texture response intensity. In this case, the initial channel weights corresponding to the initial brightness image with the strongest texture response intensity can be selected as the target channel weight reassembly corresponding to the illumination spectrum. Alternatively, the initial channel weights corresponding to the initial brightness image with the strongest texture response intensity can be selected as the weight base. Based on the texture response intensity and initial channel weights corresponding to each initial brightness image, the initial channel weights used as the weight base are adjusted, and the adjusted set of initial channel weights is used as the target channel weight reassembly.

[0138] The above technical solution obtains the target channel weights based on the texture effect of each initial brightness image, which can make the weight values ​​of each channel of the endoscopic image as close as possible to the optimal texture feature extraction effect, and is conducive to making the obtained brightness information have richer texture features.

[0139] Optionally, the texture response intensities corresponding to at least two initial brightness images are obtained, including:

[0140] For each of at least two initial brightness images

[0141] Obtain at least one region of interest in the initial brightness image;

[0142] Calculate the local edge contrast of each region of interest in the initial brightness image;

[0143] Calculate the average local edge contrast of at least one region of interest to obtain the texture response intensity corresponding to the initial brightness image.

[0144] For example, each initial brightness image can correspond to a texture response intensity. When the texture effect evaluation parameter is texture response intensity, the texture response intensity of the initial brightness image can be obtained based on the average local edge contrast of at least one region of interest (ROI) in the initial brightness image. Specifically, at least one ROI of the same size can be selected in the image texture region of the initial brightness image, for example, a 32*32 ROI. This embodiment of the invention does not specifically limit the selection conditions of the image texture region; the image texture region can be selected arbitrarily. This embodiment of the invention does not specifically limit the number of ROIs selected. When the number of ROIs selected is 1, the local edge contrast of the ROI can be used as the texture response intensity corresponding to the initial brightness image. Typically, in order to make the texture response intensities corresponding to each initial brightness image comparable, the ROIs of each initial brightness image are located in the same image position within their respective initial brightness images.

[0145] By calculating the average local edge contrast of the region of interest in the initial brightness image, the texture response intensity of each initial brightness image can be accurately obtained, so that the selection of the initial channel weights has a reference value, thereby realizing the quantification of texture effect evaluation parameters.

[0146] Optionally, for each region of interest, the local edge contrast is calculated using the following formulas (9) and (10):

[0147]

[0148] In the formula, EBCM represents the local edge contrast of the region of interest, ROI represents the region of interest in the luminance image A0, I(x,y) represents the pixel value of the pixel at coordinates (x,y) in the luminance image A0, E(x,y) represents the pixel value of the pixel in the edge image corresponding to the pixel at coordinates (x,y) in the luminance image A0, luminance image A0 is the initial luminance image corresponding to the region of interest, g represents the Gaussian filter kernel used to extract the edge image, Ω represents the region corresponding to the region of interest in the edge image, I(u,v) represents the pixel value of the pixel at coordinates (u,v) in the edge image; the pixels in the region of interest in the luminance image A0 correspond one-to-one with the pixels in the edge image.

[0149] For example, for each region of interest in any initial brightness image, the edge image corresponding to the region of interest can be obtained by extracting the region of interest in the initial brightness image through a Gaussian filter kernel. Each region of interest can be extracted by the Gaussian filter kernel to correspond to an edge image. Regarding the above formulas (12) and (13), a pixel G with coordinates (x, y) in the initial brightness image can correspond to a pixel K with coordinates (u, v) in the edge image corresponding to the region of interest where the pixel is located. Here, the coordinates (x, y) of pixel G in the initial brightness image point to the position information of pixel G relative to the initial brightness image, and the coordinates (u, v) of pixel K in the edge image point to the position information of pixel K relative to the edge image (that is, the region of interest where pixel G is located). More specifically, the position of pixel G in the initial brightness image relative to the region of interest where pixel G is located is the same as the position of pixel K relative to the edge image.

[0150] By applying Gaussian filtering to the region of interest, the impact of noise on edge detection results can be reduced, making the gray-level changes at the edges more obvious. This allows the contrast calculation to better reflect the intensity and range of changes of the edges, making the detected edge contrast more reliable.

[0151] Optionally, the first color space is the RGB color space, and the target color channel includes a first color channel and a second color channel. Based on the target channel weights corresponding to each of the at least two color channels, the endoscopic image is converted from the first color space to the second color space to obtain a target brightness image in the target brightness channel and a target color image in the target color channel, including:

[0152] The endoscopic image is converted from the first color space to the second color space using the following formulas (11), (12), and (13):

[0153]

[0154] In the formula, W c1 W represents the first color decomposition coefficient. c2 W represents the second color decomposition coefficient. b W represents the target channel weight corresponding to channel B. r Represents the target channel weight corresponding to channel R, 1-W b -W r I represents the target channel weight corresponding to channel G. r (x,y) represents the pixel value of the pixel at coordinates (x,y) in the R channel of the endoscopic image. g (x,y) represents the pixel value of the pixel at coordinates (x,y) in the G channel of the endoscopic image. b (x,y) represents the pixel value of the pixel at coordinates (x,y) in the B channel of the endoscopic image. Y (x,y) represents the brightness value of the pixel at coordinates (x,y) in the target brightness image. c1 (x,y) represents the color value of the pixel at coordinates (x,y) in the first color image under the first color channel. c2 (x,y) represents the color value of the pixel at coordinates (x,y) in the second color image under the second color channel, wherein the target color image includes the first color image and the second color image.

[0155] Since the human eye is much more sensitive to brightness than to color, the above technical solution can prioritize the accuracy of brightness information, reduce mutual interference between colors, improve the accuracy of color reproduction, and thus improve the visual effect of the acquired enhanced blood vessel image.

[0156] Optionally, the first color space is the RGB space;

[0157] When the wavelength corresponding to the principal component light is located on the short-wavelength side: W b >W g >W r ;

[0158] When the wavelength corresponding to the principal component light is located on the longer wavelength side: W b <W g <W r ;

[0159] Among them, W b +W g +W r =1, W b W represents the target channel weight corresponding to channel B, and Wg represents the target channel weight corresponding to channel G. r The target channel weight is the value corresponding to channel R.

[0160] For example, wavelengths can be categorized into short-wavelength and long-wavelength based on their numerical value. In practical applications, wavelengths smaller than a first wavelength threshold (e.g., 450 nm) can be designated as short-wavelength (i.e., located on the short-wavelength side), and wavelengths greater than or equal to a second wavelength threshold (e.g., 600 nm) can be designated as long-wavelength (i.e., located on the long-wavelength side). Alternatively, long-wavelength and short-wavelength ranges can be preset, and the wavelength's location on the short-wavelength or long-wavelength side can be determined based on the wavelength range it falls into. When the wavelength corresponding to the principal component light is on the short-wavelength side, the main color of the target object is distributed within the B channel. Consequently, the target channel weight value corresponding to the B channel is the largest, the target channel weight value corresponding to the R channel is the smallest, and the target channel weight value corresponding to the G channel is between the two. Conversely, when the wavelength corresponding to the principal component light is on the long-wavelength side, the main color of the target object is distributed within the R channel. Consequently, the target channel weight value corresponding to the R channel is the largest, the target channel weight value corresponding to the B channel is the smallest, and the target channel weight value corresponding to the G channel is between the two.

[0161] The above technical solution can set the target channel weight of each color channel according to the wavelength of the principal component light, which is beneficial to retain the main information of the tissue to be observed in the endoscopic image more completely, and helps to suppress background information (most of which are image noise or unimportant tissue) in the endoscopic image.

[0162] Optionally, before performing image reconstruction based on the target image to obtain a blood vessel enhancement image, the method further includes:

[0163] A target color image is color-enhanced to obtain an enhanced color image, wherein the target color channel includes a first color channel and a second color channel, the target color image includes a first color image under the first color channel and a second color image under the second color channel; the enhanced color image includes an enhanced first color image and an enhanced second color image.

[0164] Structural edge features are extracted from the endoscope image based on the target brightness image to obtain an initial structural edge image;

[0165] Structural edge enhancement is performed on the initial structural edge image to obtain the enhanced structural edge image;

[0166] The target image also includes an enhanced color image, an enhanced structural edge image, and a target brightness image;

[0167] Among them, image reconstruction based on the target image to obtain an enhanced blood vessel image includes:

[0168] The first color decomposition coefficient W is calculated using the following formulas (14) and (15).c1 Second color decomposition coefficient W c2 :

[0169]

[0170] The target image is subjected to an inverse linear transformation using the following formula (16):

[0171]

[0172] in, This represents the pixel value of the pixel at coordinates (x, y) in the R channel of a blood vessel enhancement image. This represents the pixel value of the pixel at coordinates (x, y) in the G channel of an enhanced blood vessel image. W represents the pixel value of the pixel at coordinates (x, y) in the B channel of a blood vessel enhancement image. b W represents the target channel weight corresponding to channel B. r Represents the target channel weight corresponding to channel R, 1-W b -W r I represents the target channel weight corresponding to channel G. Y (x,y) represents the brightness value of the pixel at coordinates (x,y) in the target brightness image. This represents the texture intensity of the pixel at coordinates (x, y) in the enhanced texture image. This represents the pixel value of the pixel at coordinates (x, y) in the enhanced structural edge image. This represents the color value of the pixel at coordinates (x, y) in the enhanced first color image. This represents the color value of the pixel at coordinates (x, y) in the enhanced second color image.

[0173] For example, the target image may include an enhanced color image, an enhanced structural edge image, an enhanced texture image, and a target brightness image. Specifically, the target color image can be enhanced using methods such as contrast stretching to obtain the enhanced color image. Structural edge features can be extracted from the target brightness image based on an edge detection algorithm, and the extracted structural edge features can be enhanced to obtain the structural edge image. The operations of performing contrast stretching on the target color image, extracting structural edge features on the target brightness image, and enhancing the initial structural edge image can refer to the aforementioned technical solutions and will not be repeated here. For each image contained in the target image, the color space of the image can be converted to RGB space using formula (11) to obtain the enhanced blood vessel image. In this process, the corresponding channel of the image can be converted to R channel, G channel, and B channel, and the parameters of the pixel points of the image (such as texture intensity, color value, etc.) are converted to parameters under R channel, G channel, and B channel (such as the pixel values ​​of each pixel point under RGB space).

[0174] Color enhancement of the target color image helps to effectively distinguish blood vessels from surrounding tissues visually. Structural edge enhancement of the initial structural edge image helps to improve the overall image clarity, resulting in a better visual effect for the final enhanced blood vessel image.

[0175] It should be noted that the above formula (16) can be adaptively modified according to the actual image contained in the target image. For example, if the target image does not contain the enhanced structural edge image, that is, if the structural edges in the endoscopic image are not enhanced, the above formula (16) will be modified accordingly. It can be replaced with If the target object does not contain an enhanced color image, then in the above formula (16) Each can be replaced with I c1 (x,y), I C2 (x,y).

[0176] Please see Figure 5 The diagram shown is a schematic block diagram of an endoscopic device 500 according to an embodiment of the present invention. According to another aspect of the present invention, an endoscopic device 500 is also provided, including a processor 510 and a memory 520. The memory 520 stores computer program instructions, which, when executed by the processor 610, are used to perform the vascular enhancement method described in any of the above embodiments.

[0177] According to another aspect of the present invention, a storage medium is also provided, on which program instructions are stored, which, when executed, are used to perform the vascular enhancement method described in any of the above embodiments. The storage medium may, for example, include a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0178] According to another aspect of the present invention, a computer program product is also provided, which executes the vascular enhancement method described in any of the above embodiments when running.

[0179] Those skilled in the art can understand the specific implementation schemes of the above-mentioned endoscopic devices, storage media, and computer program products by reading the relevant description of the vascular enhancement method above, and for the sake of brevity, they will not be described in detail here.

[0180] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0181] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0182] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0183] The above description is merely a specific embodiment of the present invention or an explanation of that embodiment. The scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for enhancing blood vessels, characterized in that, include: Acquire endoscopic images; The endoscopic image is decomposed to obtain the target brightness image under the target brightness channel and the target color image under the target color channel; Based on the target brightness image and the target color image, extract the vascular texture features in the endoscopic image to obtain an initial texture image, and perform texture enhancement on the initial texture image to obtain an enhanced texture image; Image reconstruction is performed based on the target image to obtain an enhanced blood vessel image, wherein the target image includes the enhanced texture image.

2. The method according to claim 1, characterized in that, The step of extracting vascular texture features from the endoscopic image based on the target brightness image and the target color image includes: The target brightness image and the target color image are jointly filtered based on a brightness filter kernel and a color filter kernel to extract the vascular texture features in the endoscopic image.

3. The method according to claim 2, characterized in that, The target color channel includes a first color channel and a second color channel, and the target color image includes a first color image under the first color channel and a second color image under the second color channel; The joint filtering operation is performed using the following formula: Among them, I t (x,y) represents the texture intensity of the pixel with coordinates (x,y) in the initial texture image, O represents the neighborhood of the pixel with coordinates (x,y), and I Y(center) I represents the brightness value of the center pixel of neighborhood O in the target brightness image. c1(center) I represents the color value of the center pixel of neighborhood O in the first color image. c2(center) I represents the color value of the center pixel of neighborhood O in the second color image. Y (i,j) represents the brightness value of the pixel at coordinates (i,j) in the target brightness image. c1 (i,j) represents the color value of the pixel at coordinates (i,j) in the first color image. c2 (i,j) represents the color value of the pixel at coordinates (i,j) in the second color image; G1 represents the brightness filtering kernel, G c The color filter kernel is represented by δ1, and the standard deviation of the luminance filter kernel is δ. c W is the standard deviation of the color filter kernel. xy This represents the normalization coefficient of the filter kernel.

4. The method according to claim 1, characterized in that, The process of enhancing the initial texture image to obtain the enhanced texture image includes: Texture enhancement is achieved by performing a global nonlinear transformation on the initial texture image using the following formula: In the formula, I represents the texture intensity of the pixel with coordinates (x, y) in the enhanced texture image. t (x,y) represents the texture intensity of the pixel with coordinates (x,y) in the initial texture image, a represents the first preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the first scale edge and the lower limit of the preset texture intensity range corresponding to the second scale edge, b represents the second preset texture intensity, which is the upper limit of the preset texture intensity range corresponding to the second scale edge and the lower limit of the preset texture intensity range corresponding to the third scale edge, c represents the maximum texture intensity in the initial texture image, f1 represents the preset mapping value corresponding to the first preset texture intensity a, f2 represents the preset mapping value corresponding to the second preset texture intensity b, and f3 represents the preset mapping value corresponding to the maximum texture intensity c. The edge types of blood vessels are divided into first-scale edges, second-scale edges and third-scale edges according to a preset size range, and the upper or lower limit of the size range corresponding to the first-scale edge, the second-scale edge and the third-scale edge increases sequentially.

5. The method according to claim 1, characterized in that, Prior to performing image reconstruction based on the target image to obtain a blood vessel enhancement image, the method further includes: The target color image is color enhanced to obtain an enhanced color image; The target image also includes the enhanced color image.

6. The method according to claim 5, characterized in that, The step of color enhancement of the target color image to obtain an enhanced color image includes: The color values ​​of each pixel in the target color image are input into a preset blood vessel mask model to obtain a blood vessel mask image; The target color image is contrast-stretched based on the blood vessel mask image to obtain an enhanced color image; The blood vessel mask model is used to determine whether the pixel corresponding to the input color value belongs to a blood vessel.

7. The method according to claim 6, characterized in that, The preset blood vessel mask model is constructed in the following way: Acquire the first sample endoscopic image; Multiple blood vessel color sample values ​​are extracted from the first sample endoscopic image. The blood vessel color sample values ​​are the color values ​​of the pixels belonging to the blood vessels in the target color channel. Gaussian fitting is performed on the multiple blood vessel color sample values ​​to obtain the preset blood vessel mask model.

8. The method according to claim 7, characterized in that, The target color channel includes a first color channel and a second color channel, and the preset blood vessel mask model is represented by the following formula: In the formula, Indicates the first standard color value. P(I) represents the second standard color value. c1, I c2 ) indicates that the color value in the first color channel is I. c1 And the color value in the second color channel is I. c2 The similarity between pixels and blood vessels, δ m p0 is the standard deviation of the preset blood vessel mask model, p0 is the preset minimum similarity threshold, and I m This indicates that the color value in the first color channel is I. c1 And the color value in the second color channel is I. c2 The mask value of the pixels; Wherein, the first standard color value Second standard color value and the standard deviation δ m Determined through the Gaussian fitting operation.

9. The method according to any one of claims 1-8, characterized in that, Prior to performing image reconstruction based on the target image to obtain a blood vessel enhancement image, the method further includes: Based on the target brightness image, structural edge features are extracted from the endoscope image to obtain an initial structural edge image; The initial structural edge image is enhanced to obtain an enhanced structural edge image; The target image also includes the enhanced structural edge image.

10. The method according to any one of claims 1-8, characterized in that, The step of decomposing the endoscopic image to obtain a target brightness image in the target brightness channel and a target color image in the target color channel includes: Determine the illumination spectrum of the illumination light emitted by the endoscope system when acquiring the endoscopic images; The endoscopic image is decomposed based on the illumination spectrum to obtain the target brightness image in the target brightness channel and the target color image in the target color channel.

11. The method according to claim 10, characterized in that, The endoscopic image is an image in a first color space, which includes at least two color channels. The step of decomposing the endoscopic image based on the illumination spectrum to obtain a target brightness image in the target brightness channel and a target color image in the target color channel includes: Based on the illumination spectrum, a corresponding target channel weight is assigned to each of the at least two color channels, wherein the target channel weight of the color channel corresponding to the principal component light in the illumination spectrum is greater than the target channel weight of other color channels. Based on the target channel weight corresponding to each of the at least two color channels, the endoscopic image is converted from the first color space to the second color space to obtain the target brightness image under the target brightness channel and the target color image under the target color channel; The second color space includes the target luminance channel and the target color channel.

12. The method according to claim 11, characterized in that, Before assigning a corresponding target channel weight to each of the at least two color channels based on the illumination spectrum, the method further includes: For the illumination spectrum, the following weighting operation is performed to determine the target channel weighting reassembly corresponding to the illumination spectrum: Based on the illumination spectrum, at least two sets of initial channel weights are preset, wherein each set of initial channel weights includes initial channel weights that correspond one-to-one with the at least two color channels, and the initial channel weights included in any two sets of initial channel weights are not completely the same. Acquire the second sample endoscopic image; The second sample endoscopic image is decomposed according to the at least two sets of initial channel weights to obtain at least two initial brightness images under the target brightness channel, wherein the at least two initial brightness images correspond one-to-one with the at least two sets of initial channel weights; Based on the texture effect evaluation parameters of each of the initial brightness images, a set of initial channel weights is selected from the at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum, wherein the target channel weight reassembly includes the target channel weight corresponding to each of the at least two color channels.

13. The method according to claim 12, characterized in that, The texture effect evaluation parameters include texture response intensity. The step of selecting one set of initial channel weights from the at least two sets of initial channel weights as the target channel weight reassembly corresponding to the illumination spectrum based on the texture effect evaluation parameters of each initial brightness image includes: Obtain the texture response intensity corresponding to each of the at least two initial brightness images; The initial channel weights corresponding to the initial brightness image with the strongest texture response intensity are used as the target channel weights corresponding to the illumination spectrum.

14. The method according to claim 13, characterized in that, The step of obtaining the texture response intensity corresponding to each of the at least two initial brightness images includes: For each of the at least two initial brightness images Obtain at least one region of interest in the initial brightness image; Calculate the local edge contrast of each region of interest in the initial brightness image; Calculate the average local edge contrast of the at least one region of interest to obtain the texture response intensity corresponding to the initial brightness image.

15. The method according to claim 11, characterized in that, The first color space is the RGB space; When the wavelength corresponding to the principal component light is located on the short-wavelength side: W b >W g >W r ; When the wavelength corresponding to the principal component light is located on the long-wavelength side: W b <W g <W r ; Among them, W b +W g +W r =1, W b W represents the target channel weight corresponding to channel B. g W represents the target channel weight corresponding to channel G. r The target channel weight is the value corresponding to channel R.

16. An endoscopic device, comprising a processor and a memory, characterized in that, The memory stores computer program instructions, which, when executed by the processor, are used to perform the vascular enhancement method as described in any one of claims 1-15.

17. A storage medium storing program instructions, characterized in that, The program instructions, when executed, are used to perform the vascular enhancement method as described in any one of claims 1-15.

18. A computer program product comprising computer program instructions, characterized in that, The computer program instructions, when executed, are used to perform the vascular enhancement method as described in any one of claims 1-15.