Endoscope image quality evaluation method and device based on multiple dimensions, equipment and medium

Abstract

The embodiment of the application discloses a kind of based on multi-dimension's endoscope image quality evaluation method, device, equipment and medium.The present application relates to image processing technical field.It includes: the endoscope image obtained is carried out gray processing to obtain endoscope gray image, and its brightness is calculated to obtain brightness feature;Endoscope image and endoscope gray image are carried out contrast estimation to obtain global contrast feature and local contrast feature;Endoscope image is carried out color space conversion to calculate preset value to obtain color feature;GGD model is used to quantize the MSCN coefficient of endoscope gray image to obtain naturalness feature;After endoscope image is carried out denoising processing, structure similarity is calculated to obtain noise feature;Brightness feature, contrast feature, color feature, naturalness feature and noise feature are input into image quality evaluation model to carry out quality evaluation to obtain quality score.The embodiment of the application can improve the performance and accuracy of endoscope image quality evaluation.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, device, and medium for evaluating the quality of endoscopic images based on multiple dimensions. Background Technology

[0002] Existing image quality assessment methods are divided into subjective and objective methods. Subjective assessment methods involve observers rating image quality; this approach is relatively reliable and accurate but easily affected by experimental conditions and the observer's subjective factors. Objective assessment methods utilize mathematical and engineering methods to measure images, offering advantages such as simplicity, real-time performance, repeatability, and ease of integration. Based on the degree of dependence on the distortion-free original image, these methods are categorized into three types, from high to low: full-reference, partial-reference, and no-reference. Full-reference methods require complete reference image information, while partial-reference methods only require a portion of the reference image information. However, in real-world scenarios, distortion-free original images are difficult to obtain; therefore, no-reference image quality assessment methods better meet practical needs.

[0003] In the medical field, due to the inherent uncertainties of endoscopic imaging, the quality of endoscopic images varies considerably. Poor image quality can negatively impact disease diagnosis and treatment, making image quality assessment crucial. Currently, there are few algorithms specifically designed for evaluating endoscopic image quality. Because it's difficult to simultaneously acquire high-quality and low-quality images during endoscopic image acquisition, quality assessment often relies on no-reference image quality assessment methods. However, mainstream no-reference image quality assessment methods primarily target natural scene images and computer-generated images, failing to fully exploit the distortion characteristics of endoscopic images and thus exhibiting less than ideal performance in evaluating their quality. Therefore, existing techniques suffer from the inability to accurately assess the quality of endoscopic images. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and medium for evaluating the quality of endoscopic images based on multiple dimensions, aiming to solve the problem that existing methods cannot accurately evaluate the quality of endoscopic images.

[0005] In a first aspect, embodiments of the present invention provide a method for evaluating the quality of endoscopic images based on multiple dimensions, comprising:

[0006] Acquire an endoscope image, perform grayscale processing on the endoscope image to obtain an endoscope grayscale image, and calculate the brightness of the endoscope grayscale image to obtain brightness features;

[0007] Contrast estimation is performed on the endoscopic image and the endoscopic grayscale image to obtain global contrast features and local contrast features;

[0008] The endoscopic image is converted to a color space to calculate preset values ​​and obtain color features;

[0009] The MSCN coefficients of the endoscope grayscale image are calculated, and the naturalness feature is obtained by quantizing the MSCN coefficients using the GGD model.

[0010] The endoscopic image is denoised to obtain a noise-free endoscopic image, and the noise features are obtained by calculating the structural similarity between the noise-free endoscopic image and the endoscopic image based on the SSIM method.

[0011] The brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature are input into the image quality evaluation model to obtain a quality score. The image quality evaluation model is obtained by training a support vector machine regression model using a training dataset.

[0012] Secondly, embodiments of the present invention also provide a multi-dimensional endoscopic image quality evaluation device, comprising:

[0013] The processing and computing unit is used to acquire endoscopic images, perform grayscale processing on the endoscopic images to obtain endoscopic grayscale images, and calculate the brightness of the endoscopic grayscale images to obtain brightness features;

[0014] The estimation unit is used to perform contrast estimation on the endoscope image and the endoscope grayscale image to obtain global contrast features and local contrast features;

[0015] A conversion calculation unit is used to perform color space conversion on the endoscopic image to calculate preset values ​​to obtain color features;

[0016] The computational quantization unit is used to calculate the MSCN coefficients of the endoscope grayscale image and quantize the MSCN coefficients using the GGD model to obtain the naturalness feature.

[0017] The denoising calculation unit is used to denoise the endoscope image to obtain a noise-free endoscope image, and calculate the structural similarity between the noise-free endoscope image and the endoscope image based on the SSIM method to obtain noise features;

[0018] The quality evaluation unit is used to input the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature into the image quality evaluation model to evaluate the quality and obtain a quality score. The image quality evaluation model is obtained by training a support vector machine regression model using a training dataset.

[0019] Thirdly, embodiments of the present invention also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0020] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0021] This invention provides a method, apparatus, device, and medium for evaluating the quality of endoscopic images based on multiple dimensions. The method includes: acquiring an endoscope image, performing grayscale processing on the endoscope image to obtain an endoscope grayscale image, and calculating the brightness of the endoscope grayscale image to obtain a brightness feature; performing contrast estimation on the endoscope image and the endoscope grayscale image to obtain global contrast features and local contrast features; performing color space conversion on the endoscope image to calculate a preset value to obtain a color feature; calculating the MSCN coefficient of the endoscope grayscale image and quantizing the MSCN coefficient using a GGD model to obtain a naturalness feature; performing denoising processing on the endoscope image to obtain a noise-free endoscope image, and calculating the structural similarity between the noise-free endoscope image and the endoscope image based on the SSIM method to obtain a noise feature; inputting the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature into an image quality evaluation model to perform quality evaluation and obtain a quality score, wherein the image quality evaluation model is obtained by training a support vector machine regression model using a training dataset. The technical solution of this invention uses a training dataset to train a support vector machine regression model to obtain an image quality evaluation model, which can improve the generalization of the image quality evaluation model. By inputting multi-dimensional features corresponding to endoscopic images into the image quality evaluation model for quality evaluation, a quality score is obtained, which fully characterizes the distortion characteristics of endoscopic images and can improve the performance and accuracy of endoscopic image quality evaluation. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating a multi-dimensional endoscopic image quality evaluation method provided in an embodiment of the present invention;

[0024] Figure 2A schematic diagram of a sub-process of a multi-dimensional endoscopic image quality evaluation method provided in an embodiment of the present invention;

[0025] Figure 3 A schematic block diagram of a multi-dimensional endoscopic image quality evaluation device provided for embodiments of the present invention; and

[0026] Figure 4 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0029] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0030] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0031] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0032] Please see Figure 1 , Figure 1This is a flowchart illustrating a multi-dimensional endoscopic image quality assessment method provided in an embodiment of the present invention. The multi-dimensional endoscopic image quality assessment method of this invention can be applied to support vector machine regression models. For example, it can be implemented through software programs configured corresponding to the support vector machine regression model, thereby improving the performance and accuracy of endoscopic image quality assessment. Figure 1 As shown, the method includes the following steps S110-S160.

[0033] S110. Acquire an endoscope image, perform grayscale processing on the endoscope image to obtain an endoscope grayscale image, and calculate the brightness of the endoscope grayscale image to obtain brightness features.

[0034] In this embodiment of the invention, an endoscope image is acquired, and the endoscope image is processed using a preset grayscale function to obtain an initial endoscope grayscale image, wherein the preset grayscale function is the rgb2gray function in MATLAB; the endoscope grayscale image is determined based on the initial endoscope grayscale image and a preset multiplier, wherein the preset multiplier is M = {1 / 8, 1 / 6, 1 / 4, 1 / 2, 2, 4, 6, 8}; the brightness of the endoscope grayscale image is calculated using the information entropy formula to obtain multiple brightness features, wherein the information entropy formula is as shown in formula (1), in formula (1), P j M i This represents the probability that a pixel with value j appears in the i-th endoscopic grayscale image. For ease of understanding, assume that the initial endoscopic grayscale image is G0, and the endoscopic grayscale image G... i G i =G o ·M i i = 1, 2, ..., 8, M i M1 is the i-th multiplier used for weight balancing. Understandably, if i is 1, then M1 is 1 / 8; if i is 2, then M2 is 1 / 6, and so on. It should be noted that in this embodiment, the information entropy formula is used to calculate the brightness of the endoscopic grayscale image because information entropy is widely used in image quality assessment to measure image details; therefore, information entropy E is used in this embodiment. Mi To characterize its brightness maintenance capability and improve the accuracy of endoscopic image quality assessment, the brightness feature is understood to be represented as F. B ={EM1,EM2,…,EM8}. It should also be noted that, in this embodiment, the endoscopic image is the original distorted image.

[0035]

[0036] S120. Perform contrast estimation on the endoscope image and the endoscope grayscale image to obtain global contrast features and local contrast features.

[0037] In this embodiment of the invention, the Minkowski distance formula is used to estimate the overall contrast of the endoscopic image to obtain global contrast features; specifically, assuming the global contrast is Cg, Cg is as shown in formula (2), where K represents the number of pixels in the endoscopic image, and I k I represents the k-th pixel in the endoscopic image I; p The p-th power represents the endoscopic image. It should be noted that different values ​​of p may increase the contrast weight to varying degrees; therefore, p is set to {1 / 8, 1 / 6, 1 / 4, 1 / 2, 2, 4, 6, 8} for weight balancing. q controls the degree of deviation of the calculated data from the center, and is set to 4. Since p has 8 different values, there are 8 global contrast features, which are simply labeled as FCg = {C...} g1 C g2 ,…,C g8}

[0038]

[0039] Further, local contrast estimation is performed on the endoscopic image and the endoscopic grayscale image to obtain a first local contrast feature and a second local contrast feature, and the first local contrast feature and the second local contrast feature are used as local contrast features. Specifically, the first local contrast feature is obtained by performing local contrast estimation on the endoscopic image using the contrast energy formula, wherein the contrast energy formula is as shown in formula (3), in formula (3), CE f For the first local contrast feature, f∈{gr, yb, rg} represents each color channel of the endoscopic image I, where gr=0.299R+0.587G+0.114B, yb=0.5(R+G)-B, rg=RG, and R, G and B represent each component in each color channel; f v f h These are the vertical and horizontal second derivatives of the Gaussian function, respectively, and α is used to calculate Z(I). f The maximum value of ) is γ, which is the contrast gain; φf is used to threshold the noise in color channel f. Finally, the first local contrast feature is obtained as FC. la ={CE gr CE yb CE rg}

[0040]

[0041] Furthermore, the second local contrast feature is obtained by extracting the ULBP features of the endoscope grayscale image using the ULBP formula, where the ULBP formula is shown in formula (4). In formula (4), This indicates a rotation-invariant LBP mode. Here, P is the number of neighbors considered, R is the radius of the considered neighbors, and s(.) is a sign function representing the relationship between two pixels. When vi ≥ vc, s(vi-vc) = 1; otherwise, s(vi-vc) = 0. vi is the value of the i-th neighbor of the center pixel in the endoscopic grayscale image Go. In this embodiment, P and R are set to 8 and 1 respectively. Thus, formula (5) can be obtained. Through formula (5), one non-uniform mode and nine uniform modes can be obtained. These ten local contrast features are used as the second local contrast features and labeled as F. Clb ={UL0,UL1,…,UL9}. It should be noted that in this embodiment, the ULBP operator considers pixel v. c And its relationship with adjacent pixels in a local region. It should also be noted that, in this embodiment, the contrast of the endoscopic image can be fully characterized by the global contrast feature, the first local contrast feature, and the second local contrast feature.

[0042]

[0043]

[0044] S130. Perform color space conversion on the endoscopic image to calculate preset values ​​and obtain color features. In this embodiment of the invention, due to the high correlation between the RGB three color channels of the endoscopic image, it is not suitable for color feature extraction. Therefore, the R, G, and B color channels in the endoscopic image are first converted to obtain opposing-color space components, wherein the opposing-color space components are... For each opposing color space component, the mean M, standard deviation D, and skewness S are calculated as statistics to obtain the color feature. The formulas for calculating the mean M, standard deviation D, and skewness S are shown in formulas (6)-(7). The color feature F can be obtained through formulas (6)-(7). Col ={M1,D1,S1,M2,D2,S2,M3,D3,S3}.

[0045] Mk=Avg(O k ),k∈{1,2,3} (6)

[0046]

[0047]

[0048] S140. Calculate the MSCN coefficients of the endoscope grayscale image, and use the GGD model to quantize the MSCN coefficients to obtain the naturalness feature.

[0049] In this embodiment of the invention, the MSCN coefficient of the endoscopic grayscale image It can be represented by formula (9), where i and j represent coordinates, G0 represents the grayscale image of the endoscope, and C is a constant used to avoid the denominator of formula (9) being 0. The local mean is represented by formula (10), and the standard deviation of the pixels is represented by formula (11). In formulas (9) and (10), ω = {ω p,q |p=-P,...,P;q=-Q,...Q} is a two-dimensional circular Gaussian weighted function. It should be noted that in this implementation, a zero-mean Gaussian distribution (GGD) model is used to quantify the MSCN coefficient distribution. The GGD model with a mean of 0 is shown in formula (12), where, When α > 0 Parameters τ and σ 2 The shape and variance of the distribution are controlled separately, with x being G0. Therefore, the naturalness feature is labeled F. Na ={τ,σ 2}

[0050]

[0051]

[0052]

[0053]

[0054] S150. The endoscopic image is denoised to obtain a noise-free endoscopic image, and the noise features are obtained by calculating the structural similarity between the noise-free endoscopic image and the endoscopic image based on the SSIM method.

[0055] In this embodiment of the invention, the endoscopic images often contain noise; therefore, accurately estimating the noise level is extremely helpful for image quality evaluation. In this embodiment, a Gaussian low-pass filter is used to process the noisy endoscopic image to obtain a noise-free image. Based on the SSIM method, the structural similarity between the noise-free endoscopic image and the endoscopic image is calculated to obtain noise features. The calculation formula for the SSIM method is F. No =φ(x,y), where F NoLet x be the noise feature, y be the noisy endoscope image, and φ(.) be the structural similarity calculation function.

[0056] S160. Input the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature into the image quality evaluation model to perform quality evaluation and obtain a quality score. The image quality evaluation model is obtained by training a support vector machine regression model using a training dataset.

[0057] In this embodiment of the invention, multi-dimensional features corresponding to the endoscopic image are input into an image quality evaluation model to obtain a quality score. The multi-dimensional features include brightness features, global contrast features, local contrast features, color features, naturalness features, and noise features, which fully characterize the distortion features of the endoscopic image and improve the performance and accuracy of endoscopic image quality evaluation. It should be noted that in this embodiment, the image quality evaluation model is trained using a Support Vector Regression (SVR) model on a training dataset. Furthermore, in this embodiment, a one-dimensional quality score is obtained by inputting 41 dimensions—8 brightness features, 8 global contrast features, 3 first local contrast features, 10 second local contrast features, 9 color features, 2 naturalness features, and 1 noise feature—into the trained SVR model.

[0058] Please see Figure 2 The image quality assessment model is obtained by training the support vector machine regression model using the training dataset, specifically including the following steps S161-S163:

[0059] S161. For each training batch of images in the training dataset, the brightness feature, global contrast feature, local contrast feature, color feature, naturalness feature, and noise feature corresponding to the training batch of images are input into the support vector machine regression model to output a predicted quality score.

[0060] S162. Calculate the loss value using a loss function based on the predicted quality score and the label quality score in the training dataset.

[0061] S163. The support vector machine regression model is iteratively updated according to the loss value until a preset number of training batches is reached to obtain the image quality evaluation model.

[0062] In this embodiment of the invention, before training the support vector machine regression model, 2400 distorted endoscopic images were collected, and 18 volunteers subjectively rated each image. The average of the ratings was used as the quality score for each image. The 2400 images were then divided into a training set of 1800 images and a test set of 600 images, i.e., the ratio of the training dataset to the test dataset was 3:1. Understandably, in this embodiment, the training dataset is used for the training phase of the support vector machine regression model, and the test dataset is used to test the optimized support vector machine regression model. In other embodiments, the number of images in the training dataset and the test dataset is not specifically limited.

[0063] Further, during the training of the quality assessment model, the brightness features, global contrast features, local contrast features, color features, naturalness features, and noise features corresponding to the training batch images are input into the support vector machine regression model to output a predicted quality score. A loss value is calculated using a loss function based on the predicted quality score and the label quality scores in the training dataset, where the loss function is a mean squared error loss function. The support vector machine regression model is iteratively updated based on the loss value until a preset number of training batches is reached to obtain the image quality assessment model. Specifically, it is determined whether the loss value is less than the previous loss value. If the loss value is not less than the previous loss value, it indicates that the loss value remains stable, and the trained support vector machine regression model is used as the image quality assessment model. Conversely, if the loss value is less than the previous loss value, it indicates that the loss value is still decreasing, and the network parameters are further optimized, and step S161 is returned to continue training the support vector machine regression model.

[0064] Furthermore, in this embodiment of the invention, after reaching a preset number of training batches, the brightness feature, global contrast feature, local contrast feature, color feature, naturalness feature, and noise feature corresponding to each test image in the test dataset are input into the trained quality evaluation model to obtain a test quality score. An index value is calculated based on the test quality score and the label quality score in the test dataset, wherein the index value is the SRCC coefficient and the PLCC coefficient. Understandably, the closer the SRCC coefficient and PLCC coefficient are to 1, the better the performance of the image quality evaluation model and the more accurate the image quality evaluation. It should be noted that in this embodiment, it has been verified that the SRCC value is above 0.82 and the PLCC value is above 0.83, thus indicating that the image quality evaluation model in this embodiment has a good evaluation effect on endoscopic images.

[0065] Figure 3 This is a schematic block diagram of a multi-dimensional endoscopic image quality evaluation device 200 provided in an embodiment of the present invention. Figure 3 As shown, corresponding to the above-described multi-dimensional endoscopic image quality evaluation method, the present invention also provides a multi-dimensional endoscopic image quality evaluation device 200. This multi-dimensional endoscopic image quality evaluation device 200 includes a unit for performing the above-described multi-dimensional endoscopic image quality evaluation method, and the device can be configured in a computer device. Specifically, please refer to... Figure 3 The multi-dimensional endoscopic image quality evaluation device 200 includes a processing calculation unit 201, an estimation unit 202, a conversion calculation unit 203, a calculation quantization unit 204, a noise reduction calculation unit 205, and a quality evaluation unit 206.

[0066] The processing and calculation unit 201 is used to acquire an endoscope image, perform grayscale processing on the endoscope image to obtain an endoscope grayscale image, and calculate the brightness of the endoscope grayscale image to obtain brightness features; the estimation unit 202 is used to perform contrast estimation on the endoscope image and the endoscope grayscale image to obtain global contrast features and local contrast features; the conversion and calculation unit 203 is used to perform color space conversion on the endoscope image to calculate preset values ​​to obtain color features; the calculation and quantization unit 204 is used to calculate the MSCN coefficients of the endoscope grayscale image and use the GGD model to quantize the MSCN coefficients. The naturalness feature is obtained by quantizing the data; the denoising calculation unit 205 is used to denoise the endoscope image to obtain a noise-free endoscope image, and calculate the structural similarity between the noise-free endoscope image and the endoscope image based on the SSIM method to obtain the noise feature; the quality evaluation unit 206 is used to input the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature and the noise feature into the image quality evaluation model to perform quality evaluation and obtain a quality score, wherein the image quality evaluation model is obtained by training a support vector machine regression model using a training dataset.

[0067] In some embodiments, such as this one, the processing and calculation unit 201 includes a grayscale processing unit, a determination unit, and a first calculation unit.

[0068] The grayscale processing unit is used to process the endoscope image using a preset grayscale function to obtain an initial endoscope grayscale image; the determining unit is used to determine the endoscope grayscale image based on the initial endoscope grayscale image and a preset multiplier; the first calculation unit is used to calculate the brightness of the endoscope grayscale image using the information entropy formula to obtain multiple brightness features.

[0069] In some embodiments, such as this one, the estimation unit 202 includes a global estimation unit and a local estimation unit.

[0070] The global estimation unit is used to perform global contrast estimation on the endoscope image using the Minkowski distance formula to obtain global contrast features; the local estimation unit is used to perform local contrast estimation on the endoscope image and the endoscope grayscale image to obtain a first local contrast feature and a second local contrast feature, and uses the first local contrast feature and the second local contrast feature as local contrast features.

[0071] In some embodiments, such as this embodiment, the local estimation unit includes a first local estimation subunit and a second local estimation subunit.

[0072] The first local estimation subunit is used to perform local contrast estimation on the endoscope image using the contrast energy formula to obtain a first local contrast feature; the second local estimation subunit is used to extract the ULBP features of the endoscope grayscale image using the ULBP formula to obtain a second local contrast feature.

[0073] In some embodiments, such as this one, the conversion calculation unit 203 includes a conversion unit and a second calculation unit.

[0074] The conversion unit is used to perform color space conversion on the color channels in the endoscopic image to obtain the opposite color space components; the second calculation unit is used to calculate the average value, standard deviation, and skewness of the opposite color space components to obtain color features.

[0075] In some embodiments, such as this one, the step of training a support vector machine regression model using a training dataset to obtain an image quality evaluation model includes a first input-output unit, a third computation unit, and an iterative update unit.

[0076] The first input / output unit is used to input the brightness feature, global contrast feature, local contrast feature, color feature, naturalness feature, and noise feature corresponding to each training batch of images in the training dataset into the support vector machine regression model to output a predicted quality score; the third calculation unit is used to calculate a loss value based on the predicted quality score and the label quality score in the training dataset using a loss function; the iterative update unit is used to iteratively update the support vector machine regression model based on the loss value until a preset number of training batches is reached to obtain an image quality evaluation model.

[0077] In some embodiments, such as this one, the step of training the support vector machine regression model using the training dataset to obtain the image quality evaluation model further includes a second input-output unit and a fourth computation unit.

[0078] The second input / output unit is used to input the brightness feature, global contrast feature, local contrast feature, color feature, naturalness feature, and noise feature corresponding to each test image in the test dataset into the trained quality evaluation model to obtain a test quality score; the fourth calculation unit is used to calculate an index value based on the test quality score and the label quality score in the test dataset.

[0079] The specific implementation of the multi-dimensional endoscopic image quality evaluation device 200 in this embodiment corresponds to the multi-dimensional endoscopic image quality evaluation method described above, and will not be repeated here.

[0080] The aforementioned multi-dimensional endoscopic image quality assessment device can be implemented as a computer program, which can, for example... Figure 4 It runs on the computer device shown.

[0081] Please see Figure 4 , Figure 4 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 300 is a server. Specifically, the server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0082] See Figure 4 The computer device 300 includes a processor 302, a memory, and a network interface 305 connected via a system bus 301. The memory may include a storage medium 303 and internal memory 304.

[0083] The storage medium 303 may store an operating system 3031 and a computer program 3032. When the computer program 3032 is executed, it causes the processor 302 to execute a multi-dimensional endoscopic image quality evaluation method.

[0084] The processor 302 provides computing and control capabilities to support the operation of the entire computer device 300.

[0085] The internal memory 304 provides an environment for the operation of the computer program 3032 in the storage medium 303. When the computer program 3032 is executed by the processor 302, the processor 302 can execute a multi-dimensional endoscopic image quality evaluation method.

[0086] This network interface 305 is used for network communication with other devices. Those skilled in the art will understand that... Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 300 to which the present application is applied. The specific computer device 300 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0087] The processor 302 is used to run a computer program 3032 stored in a memory to implement the process steps of the above-described method embodiments.

[0088] It should be understood that in the embodiments of this application, the processor 302 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0089] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0090] Therefore, the present invention also provides a storage medium. This storage medium may be a computer-readable storage medium. The storage medium stores a computer program. When executed by a processor, the computer program causes the processor to perform any embodiment of the above-described multi-dimensional endoscopic image quality assessment method.

[0091] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0092] 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, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. 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.

[0093] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0094] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0096] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0097] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Since these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

[0098] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A multi-dimensional based endoscopic image quality evaluation method applied to a constructed polyp segmentation model, characterized in that, The method comprises the following steps: An endoscope image is acquired, and the endoscope image is subjected to grayscale processing to obtain an endoscope grayscale image, and the brightness of the endoscope grayscale image is calculated to obtain a brightness feature; The endoscope image and the endoscope grayscale image are subjected to contrast estimation to obtain a global contrast feature and a local contrast feature; The endoscope image is subjected to color space conversion to calculate a preset value to obtain a color feature; The MSCN coefficient of the endoscope grayscale image is calculated, and the GGD model is used to quantize the MSCN coefficient to obtain a naturalness feature; The endoscope image is subjected to denoising processing to obtain a noise-free endoscope image, and the structural similarity between the noise-free endoscope image and the endoscope image is calculated based on the SSIM method to obtain a noise feature; The brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature and the noise feature are input into an image quality evaluation model for quality evaluation to obtain a quality score, wherein the image quality evaluation model is obtained by training a support vector machine regression model using a training data set; The contrast estimation of the endoscope image and the endoscope grayscale image to obtain a global contrast feature and a local contrast feature comprises: The Minkowski distance formula is used to estimate the overall contrast of the endoscope image to obtain a global contrast feature; The contrast energy formula is used to estimate the local contrast of the endoscope image to obtain a first local contrast feature; The ULBP feature of the endoscope grayscale image is extracted by the ULBP formula to obtain a second local contrast feature, and the first local contrast feature and the second local contrast feature are taken as the local contrast feature.

2. The method of claim 1, wherein, The grayscale processing of the endoscope image to obtain an endoscope grayscale image and the calculation of the brightness of the endoscope grayscale image to obtain a brightness feature comprise: The endoscope image is subjected to grayscale processing using a preset grayscale function to obtain an initial endoscope grayscale image; The endoscope grayscale image is determined according to the initial endoscope grayscale image and a preset multiplier; The brightness of the endoscope grayscale image is calculated by the information entropy formula to obtain a plurality of brightness features.

3. The method of claim 1, wherein, The color space conversion of the endoscope image to calculate a preset value to obtain a color feature comprises: The color channels in the endoscope image are subjected to color space conversion to obtain an opponent color color space component; The mean, standard deviation and skewness of the opponent color color space component are calculated to obtain a color feature.

4. The method of claim 1, wherein, The training of a support vector machine regression model using a training data set to obtain an image quality evaluation model comprises: For each training batch image in the training data set, the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature and the noise feature corresponding to the training batch image are input into the support vector machine regression model to output a predicted quality score; The loss value is calculated by a loss function according to the predicted quality score and the label quality score in the training data set; The support vector machine regression model is iteratively updated according to the loss value until a preset training batch number is reached, so as to obtain an image quality evaluation model.

5. The method of claim 4, wherein, After the support vector machine regression model is trained by using the training data set to obtain the image quality evaluation model, the method further includes: The brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature corresponding to each test image in the test data set are input into the trained quality evaluation model to obtain a test quality score. An index value is calculated according to the test quality score and a label quality score in the test data set.

6. A multi-dimensional based endoscope image quality evaluation apparatus, characterized by, It includes: A processing calculation unit is configured to acquire an endoscope image, perform grayscale processing on the endoscope image to obtain an endoscope grayscale image, and calculate the brightness of the endoscope grayscale image to obtain a brightness feature; An estimation unit is configured to perform contrast estimation on the endoscope image and the endoscope grayscale image to obtain a global contrast feature and a local contrast feature; A conversion calculation unit is configured to perform color space conversion on the endoscope image to calculate a preset value to obtain a color feature; A calculation and quantization unit is configured to calculate the MSCN coefficient of the endoscope grayscale image, and quantize the MSCN coefficient by using a GGD model to obtain a naturalness feature; A denoising calculation unit is configured to perform denoising processing on the endoscope image to obtain a noise-free endoscope image, and calculate the structural similarity between the noise-free endoscope image and the endoscope image by using an SSIM method to obtain a noise feature; A quality evaluation unit is configured to input the brightness feature, the global contrast feature, the local contrast feature, the color feature, the naturalness feature, and the noise feature into an image quality evaluation model to perform quality evaluation and obtain a quality score, wherein the image quality evaluation model is obtained by training a support vector machine regression model by using a training data set. The estimation unit includes: An overall estimation unit is configured to perform overall contrast estimation on the endoscope image by using a Minkowski distance formula to obtain a global contrast feature; A first local estimation subunit is configured to perform local contrast estimation on the endoscope image by using a contrast energy formula to obtain a first local contrast feature; A second local estimation subunit is configured to extract ULBP features of the endoscope grayscale image by using an ULBP formula to obtain a second local contrast feature, and use the first local contrast feature and the second local contrast feature as local contrast features.

7. A computer device, comprising: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the method of any one of claims 1-5 when executing the computer program.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is executed by the processor to implement the method of any one of claims 1-5.

Images