An image quality judgment method, device, equipment and storage medium

Abstract

Embodiments of the present application disclose an image quality judgment method, device and equipment and a storage medium. The technical scheme provided by the embodiments of the present application determines original statistical information according to the temperature mean value of each original image in an original image set, performs clustering processing on the original image set based on a first image class center and a second image class center, determines the original image corresponding to the first image class center and the second image class center after the clustering processing, determines first clustering statistical information according to the temperature mean value of the original image corresponding to the first image class center, and determines second clustering statistical information according to the temperature mean value of the original image corresponding to the second image class center, and finally determines whether the original images are all obtained by an infrared camera in a thermal stable state according to the original statistical information, the first clustering statistical information and the second clustering statistical information, so as to quickly and accurately judge the image temperature measurement quality of the original image set.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to an image quality judgment method, apparatus, device, and storage medium. Background Technology

[0002] In the reconstruction process of 3D scenes, image fusion and stitching are often performed on multiple frames of infrared images captured continuously by an infrared camera to obtain a stitched image. For example, an infrared camera mounted on a drone can be used to capture multiple frames of infrared images by controlling the drone to move over the target scene. Then, the pose information between the multiple frames of infrared images is used to construct 3D scene information. An orthophoto image set is obtained from the 3D scene information and stitched together to form a 2D orthophoto image.

[0003] Because infrared cameras have a thermal stabilization time during operation, and the temperature measurement is unstable when the camera captures images during this time, it is unclear whether the infrared camera had reached a thermally stable state when the images were captured. Therefore, it is impossible to judge the temperature detection quality of the infrared images. If the original infrared images captured when the infrared camera had not reached a thermally stable state are directly used for image stitching, it is easy to cause obvious discontinuities in the stitched images after stitching due to temperature jumps, resulting in unsatisfactory image stitching effects. Summary of the Invention

[0004] This application provides an image quality judgment method, apparatus, device, and storage medium to solve the technical problem in the prior art where the thermal stability of the output infrared image is uncertain and the image stitching effect is not ideal. The method judges the thermal stability of the infrared image to determine whether temperature compensation is needed for the original image, thereby ensuring a good image stitching effect.

[0005] In a first aspect, embodiments of this application provide an image quality judgment method, including:

[0006] Determine the average temperature of each original image in the original image set, and determine the original statistical information based on the average temperature;

[0007] The original image set is clustered based on the first image class center and the second image class center to determine the original images corresponding to the first image class center and the second image class center after clustering.

[0008] First clustering statistics are determined based on the average temperature of the original image corresponding to the first image cluster center, and second clustering statistics are determined based on the average temperature of the original image corresponding to the second image cluster center.

[0009] Based on the original statistical information, the first clustering statistical information, and the second clustering statistical information, it is determined whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0010] In a second aspect, embodiments of this application provide an image quality judgment device, including an initial statistics module, an image clustering module, a clustering statistics module, and a stability judgment module, wherein:

[0011] The initial statistics module is used to determine the average temperature of each original image in the original image set, and to determine the original statistical information based on the average temperature.

[0012] The image clustering module is used to perform clustering processing on the original image set based on the first image cluster center and the second image cluster center, and to determine the original images corresponding to the first image cluster center and the second image cluster center after the clustering processing, respectively.

[0013] The clustering statistics module is used to determine first clustering statistics based on the average temperature of the original image corresponding to the first image cluster center, and to determine second clustering statistics based on the average temperature of the original image corresponding to the second image cluster center.

[0014] The stability determination module is used to determine, based on the original statistical information, the first clustering statistical information, and the second clustering statistical information, whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0015] In a third aspect, embodiments of this application provide an image quality assessment device, including: a memory and one or more processors;

[0016] The memory is used to store one or more programs;

[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the image quality judgment method as described in the first aspect.

[0018] In a fourth aspect, embodiments of this application provide a storage medium for storing computer-executable instructions, which, when executed by a computer processor, are used to perform the image quality judgment method as described in the first aspect.

[0019] This application embodiment determines the original statistical information based on the average temperature of each original image in the original image set, and performs clustering processing on the original image set based on the first image class center and the second image class center to determine the original images corresponding to the first image class center and the second image class center after clustering processing. The first clustering statistical information is determined based on the average temperature of the original images corresponding to the first image class center, and the second clustering statistical information is determined based on the average temperature of the original images corresponding to the second image class center. Finally, based on the original statistical information, the first clustering statistical information and the second clustering statistical information, it is determined whether all the original images in the original image set were captured in the thermally stable state of the infrared camera. This allows for a quick and accurate assessment of the image temperature measurement quality of the original image set, enabling a rapid determination of whether temperature compensation is needed for the original images before stitching them together. Attached Figure Description

[0020] Figure 1 This is a flowchart of an image quality judgment method provided in an embodiment of this application;

[0021] Figure 2 This is a schematic diagram illustrating the display effect of a stitched image without temperature compensation, provided in an embodiment of this application.

[0022] Figure 3 This is a schematic diagram illustrating the display effect of a target stitched image provided in an embodiment of this application;

[0023] Figure 4 This is a flowchart of another image quality judgment method provided in the embodiments of this application;

[0024] Figure 5 This is a schematic diagram of an iterative update process for class labels of an original image provided in an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of a thermal stability state determination process provided in an embodiment of this application;

[0026] Figure 7 This is a schematic diagram of the structure of an image quality judgment device provided in an embodiment of this application;

[0027] Figure 8 This is a schematic diagram of the structure of an image quality assessment device provided in an embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0029] Figure 1 A flowchart of an image quality judgment method provided in an embodiment of this application is given. The image quality judgment method provided in this embodiment of the application can be executed by an image quality judgment device, which can be implemented by hardware and / or software and integrated into an image quality judgment device.

[0030] The following description uses an image quality assessment device to perform an image quality assessment method as an example. (Reference) Figure 1 The image quality assessment method includes:

[0031] S101: Determine the average temperature of each original image in the original image set, and determine the original statistical information based on the average temperature.

[0032] The original image set comprises multiple original images, which can be captured by infrared cameras mounted on mobile devices (such as drones, unmanned vehicles, etc.). The original images captured by the infrared cameras also record corresponding temperature information and the camera's pose information. The original image set can be obtained by continuously capturing multiple frames of the scene using an infrared camera mounted on a drone. This original image set can be used to reconstruct the 3D data of the entire scene, constructing a stitched image of orthorectified infrared thermal imaging data for a large-scale scene, and obtaining global scene image information. For example, the pose information between multiple frames of original images can be used to construct 3D scene information. An orthorectified projection image set can then be obtained from the 3D scene information and stitched together to form a 2D orthorectified stitched image.

[0033] In related technologies, infrared cameras have a thermal stabilization time during operation. During this time, the temperature measurement during image capture is unstable. If the camera is captured before the thermal stabilization time has ended, the original image captured during the thermal stabilization time and the original image captured after the thermal stabilization time will have different display effects due to the difference in temperature measurement stability. Consequently, in the final stitched image, obvious discontinuities will occur due to temperature jumps, resulting in poor display quality of the stitched image.

[0034] like Figure 2 The provided illustration shows a display effect of a stitched image without temperature compensation. It assumes that a drone equipped with an infrared camera acquires the original image of the target scene, where the drone's flight path is from... Figure 2 Starting from the bottom right corner of the image, the drone flies from right to left, acquiring raw images at set time intervals or flight distances. Upon reaching the far left of the target scene, it turns and collects raw images from left to right. Each time the drone turns, it records its trajectory... Figure 2 The image is shifted upwards until the original image sequence of each location in the target scene is acquired, resulting in the original image set. Generally, adjacent original images (including original images adjacent in multiple directions such as top, bottom, left, and right) have overlapping areas.

[0035] Assuming the infrared camera begins acquiring raw images without waiting for thermal stabilization time, the stitched image obtained from stitching together the raw image set will exhibit significant temperature anomalies in corresponding areas due to the unstable temperature acquisition of the raw images captured during the thermal stabilization period. For example... Figure 2 As shown in the darker area at the bottom of the stitched image, the display effect of the stitched image is not ideal. In related technologies, to ensure the display effect of the stitched image, it is necessary to wait for the thermal stabilization time of the infrared camera to end before acquiring the original image, resulting in low work efficiency. When it is determined that the original image set has not reached a stable state, temperature compensation can be performed frame by frame on the original image by utilizing the overlapping area information of multiple original images and the temperature consistency between adjacent original images, or by utilizing the overlapping area information of multiple original images and the temperature consistency between adjacent original images, to eliminate the difference in display effect caused by the temperature measurement difference due to the thermal stabilization time of the infrared camera, thus ensuring the display effect of the final target stitched image.

[0036] Since it is unknown whether any original images in the original image set were captured within the thermal stabilization time of the infrared camera before image stitching, it is necessary to determine whether all original images in the original image set were captured within the thermal stabilization state of the infrared camera. If all original images in the original image set were captured within the thermal stabilization state of the infrared camera (in this case, all original images in the original image set were captured after the thermal stabilization time of the infrared camera), image stitching can be performed directly using the original images. However, if there are original images in the original image set captured before the infrared camera reached the thermal stabilization state (in this case, some original images in the original image set were captured within the thermal stabilization time of the infrared camera), temperature compensation needs to be performed on the original images captured within the thermal stabilization time. This will migrate the temperature level (statistical information) of the original images captured within the thermal stabilization time to a range that is the same as or similar to the temperature level (statistical information) of the original images captured after the thermal stabilization time, ensuring the image stitching effect.

[0037] For example, the temperature information of each original image in the original image set is determined. The temperature information of each original image includes the temperature value of each pixel (or partially sampled pixels) of each original image, and the average temperature of each original image is calculated (average temperature = sum of temperature values ​​of each pixel / number of pixels).

[0038] Furthermore, the raw statistical information of the original image set is determined based on the mean temperature of each original image in the original image set. This raw statistical information reflects the temperature distribution when the infrared camera acquired the initial image set. Optionally, the raw statistical information can be represented by one or more combinations of the mean, standard deviation, and variance of the mean temperature of each original image in the original image set.

[0039] S102: Cluster the original image set based on the first image class center and the second image class center, and determine the original images corresponding to the first image class center and the second image class center after the clustering process.

[0040] For example, two class centers (a first image class center and a second image class center) are set and initialized, and the original image set is clustered based on the first image class center and the second image class center to determine the original images corresponding to the first image class center and the second image class center after clustering. The clustering result of the original image set can be understood as dividing each original image into two groups of original images corresponding to the first image class center and the second image class center, respectively.

[0041] The original images can be clustered based on the mean temperature and mean index (mean image number of the original images) of the cluster centers. In one possible embodiment, the first and second image cluster centers can be initialized based on the first and last original images in the original image set, and the first and second image cluster centers can be iteratively updated during the clustering process.

[0042] In this embodiment, the temperature data of the entire original image set is divided into two initial categories, the center data of the two categories are updated, and a clustering method is used to determine whether there are two types of temperature data clusters in the original image set. If some original images were taken within the thermally stable time, there will be two types of temperature data clusters. The distribution of the two types of temperature data clusters (distribution curves) can be used to determine whether all original images in the original image set were taken when the infrared camera reached a thermally stable state. For example, the peak values ​​of the distribution of the center of the first image class and the center of the second image class can be used to determine whether all original images in the original image set were taken when the infrared camera reached a thermally stable state.

[0043] S103: Determine the first clustering statistics based on the average temperature of the original image corresponding to the first image cluster center, and determine the second clustering statistics based on the average temperature of the original image corresponding to the second image cluster center.

[0044] For example, after clustering the original image set based on the first image cluster center and the second image cluster center, the average temperature of each original image corresponding to the first image cluster center after clustering is determined, the first clustering statistics of the first image cluster center are determined based on the average temperature of these original images, and the average temperature of each original image corresponding to the second image cluster center after clustering is determined, and the second clustering statistics of the second image cluster center are determined based on the average temperature of these original images.

[0045] The first cluster statistics reflect the temperature distribution of each original image within the first image cluster center, and the second cluster statistics reflect the temperature distribution of each original image within the second image cluster center. Furthermore, the representation of the first and second cluster statistics is consistent with the original statistics; that is, the first and second cluster statistics can also be represented by one or more combinations of the mean, standard deviation, and variance of the temperature mean of each original image corresponding to the image cluster center.

[0046] S104: Based on the original statistical information, the first clustering statistical information, and the second clustering statistical information, determine whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0047] For example, the original statistical information is compared with the first cluster statistical information and the second cluster statistical information, and the comparison results are used to determine whether all the original images in the original image set were captured in the thermally stable state of the infrared camera. For example, using the kurtosis evaluation values ​​corresponding to the original statistical information, the first cluster statistical information, and the second cluster statistical information to determine whether the original images in the original image set were captured when the infrared camera reached a thermally stable state (i.e., determining whether all the original images in the original image set were captured outside the thermally stable time of the infrared camera), if there are original images captured within the thermally stable time in the original image set, there are two types of temperature data clusters in the original image set. Correspondingly, the kurtosis evaluation value after clustering is larger, which indicates that the infrared camera had not reached a thermally stable state when the original images were captured, and there are original images in the original image set whose temperature detection quality does not meet the requirements, requiring temperature compensation for the original images. However, if all the original images in the original image set were captured after the thermally stable time, there are no two types of temperature data clusters in the original image set, and correspondingly, the kurtosis evaluation value is larger, which indicates that the infrared camera had reached a thermally stable state when the original images were captured, and the temperature detection quality of all original images in the original image set has met the requirements, requiring no temperature compensation for the original images.

[0048] In one embodiment, if it is determined that all original images in the original image set were captured when the infrared camera reached a thermally stable state, the target stitched image can be obtained directly by stitching together the original images in the original image set according to their aligned positions. However, if it is determined that the original image set contains original images captured before the infrared camera reached a thermally stable state, the images to be compensated (original images captured during the thermal stabilization period) and the thermally stable images (original images captured after the thermal stabilization period) in the original image set can be determined first. Temperature compensation is then applied to the images to be compensated, and the target stitched image is obtained by stitching together the temperature-compensated images to be compensated and the thermally stable images.

[0049] For example, the image to be compensated can be determined from the original image set based on temperature abrupt change points in the original image set. A reference image for each image to be compensated can be determined from the original image set based on the positional information of the image to be compensated and the original image set. Temperature compensation is then performed on the image to be compensated based on a first mask image of the image to be compensated and the reference image. Finally, the original image set and the temperature-compensated image to be compensated are stitched together to obtain the target stitched image. Alternatively, the image to be compensated and a thermally stable image can be determined from the original image set based on temperature abrupt change points in the original image set. Reference statistical information is determined based on the temperature information of the thermally stable image. Temperature compensation is then performed on the image to be compensated based on the reference statistical information and the temperature information of the image to be compensated. Finally, the thermally stable image and the temperature-compensated image to be compensated are stitched together to obtain the target stitched image.

[0050] When determining the image to be compensated from the original image set based on temperature abrupt changes in the original image set, the process may include: generating a set of sample pairs based on second mask images corresponding to adjacent original images in the original image set, where the sample pairs in the set indicate the temperature values ​​of adjacent original images within the second mask images; determining a temperature gain set based on the sample pair set, where the temperature gains in the temperature gain set indicate the temperature gain values ​​between adjacent original images; determining an image index value from the temperature gain set, and determining the image to be compensated from the original image set based on the image index value. Alternatively, when determining the image to be compensated from the original image set based on temperature abrupt changes in the original image set, the process may include: determining a set of temperature mean ratios based on the temperature mean of each original image in the original image set, where the temperature mean ratios in the temperature mean ratio set indicate the ratio of the temperature mean values ​​of adjacent original images; determining an image index value from the temperature mean ratio set, and determining the image to be compensated from the original image set based on the image index value. The image index value reflects the temperature abrupt changes in the original image set.

[0051] For example, after obtaining the original image set, each original image in the original image set is first aligned to determine the image position of each original image in the stitched image. Then, the image to be compensated and the thermally stable image are determined from the original image set, and reference statistical information is determined based on the temperature information of the thermally stable image. Based on the reference statistical information and the temperature information of the image to be compensated, temperature compensation is performed on each image to be compensated. Then, based on the image positions of the temperature-compensated image to be compensated and the thermally stable image on the stitched image, the image to be compensated and the thermally stable image are stitched together using a fusion feathering algorithm to obtain the target stitched image. Alternatively, after obtaining the original image set, first perform alignment operations on each original image in the original image set to determine the image position of each original image in the stitched image. Then, determine the image to be compensated and the thermally stable image from the original image set, and determine the reference image for each image to be compensated. Based on the overlapping area information of the reference image and the image to be compensated (temperature information of the first mask image), perform temperature compensation on each image to be compensated frame by frame in reverse order using temperature consistency. Then, based on the image positions of the temperature-compensated image to be compensated and the thermally stable image on the stitched image, perform image stitching on the temperature-compensated image to be compensated and the thermally stable image based on the fusion feathering algorithm to obtain the target stitched image.

[0052] like Figure 3 The provided illustration shows the display effect of a target image stitching. If image stitching is performed directly based on the original image set, after aligning and blending the original images, the result will be... Figure 2 The original stitched image shown shows that the display area corresponding to the original image captured during the thermal stabilization period is significantly darker, and the stitched image exhibits obvious tonal discontinuity. The image quality assessment method provided in this solution determines that not all original images were captured under the thermally stable condition of the infrared camera. After aligning, temperature compensating, and fusing the original image set, the result is as follows: Figure 3 As shown in the target stitched image, it can be seen that the display area corresponding to the original image captured during the thermal stabilization time is basically consistent with the display area corresponding to the original image captured after the thermal stabilization time. There is no dark local display effect or temperature discontinuity as in the original stitched image. Furthermore, it does not require waiting for the infrared camera to reach thermal stabilization time before capturing the original image, resulting in higher efficiency in both original image acquisition and image stitching.

[0053] The above describes a process that determines raw statistical information based on the average temperature of each raw image in the raw image set. Then, the raw image set is clustered based on the first and second image cluster centers to determine the raw images corresponding to the first and second image cluster centers after clustering. First clustering statistical information is determined based on the average temperature of the raw images corresponding to the first image cluster centers, and second clustering statistical information is determined based on the average temperature of the raw images corresponding to the second image cluster centers. Finally, the raw statistical information, the first clustering statistical information, and the second clustering statistical information are used to determine whether all raw images in the raw image set were captured under thermally stable conditions using an infrared camera. This allows for a rapid and accurate assessment of the image temperature measurement quality of the raw image set, enabling a quick determination of whether temperature compensation is needed before stitching the raw images together.

[0054] Based on the above embodiments, Figure 4 A flowchart of another image quality assessment method provided in an embodiment of this application is given, which is a concretization of the above-described image quality assessment method. (Reference) Figure 4 The image quality assessment method includes:

[0055] S201: Determine the average temperature of each original image in the original image set, and determine the original statistical information based on the average temperature.

[0056] For example, the original statistical information provided in this embodiment includes the statistical mean and the statistical standard deviation. The statistical mean is determined based on the mean of the temperature mean of each original image in the original image set, that is, the statistical mean is the mean of the temperature mean of each original image in the original image set. The statistical standard deviation is determined based on the standard deviation of the temperature mean of each original image in the original image set, that is, the statistical standard deviation is the standard deviation of the temperature mean of each original image in the original image set.

[0057] For example, suppose the number of original images in the original image set is N, and the width and height of the final stitched image are W and H respectively. The set of position information (image position) of each original image in the final stitched image is P = {(px1, Py1), (px2, pY2), ..., (px... N Py N Obtain the original image set S0 = {I1, I2, ..., I...}. N} Calculate the mean temperature m and standard deviation temperature s of all original images in the original image set to obtain the statistical information set S1 = {(m1, s1), (m2, s2), ..., (m N s NBased on the mean temperature m of each original image in the statistical information set S1, calculate the mean SGmean and standard deviation SGstd of the mean temperature m of all original images, and obtain the original statistical information Ω1={SGmean,SGstd} corresponding to the original image set.

[0058] The formula for calculating the mean SGmean of each temperature mean m is:

[0059] Where i is the image number of the original image.

[0060] The formula for calculating the standard deviation SGstd of the mean temperature m is:

[0061] S202: Initialize the first image class center and the second image class center, and assign class labels to each original image in the original image set based on the set clustering metric. The class labels include a first class label corresponding to the first image class center and a second class label corresponding to the second image class center.

[0062] For example, the first image class center and the second image class center are initialized, wherein the initial first image class center can be set according to the first original image in the original image set, and the initial second image class center can be set according to the last original image in the original image set.

[0063] In one embodiment, the clustering metrics for clustering the original image set include the mean temperature of the class centers and the mean index of the class centers. Correspondingly, the image class centers can be represented by the mean temperature of the class centers and the mean index of the class centers.

[0064] For example, the mean temperature m1 and the image index value index1 (e.g., index1 = 1) of the first original image in the original image set are used as the initial mean temperature value mean_mean1 and mean index value mean_index1 of the first image class center, respectively, to obtain the first image class center C1 = {mean_mean1, mean_index1}, and the mean temperature m1 of the last original image in the original image set is used as the initial mean temperature value mean_mean1 and mean index value mean_index1 of the first image class center, respectively. N and image index value N (e.g., index) N =N) are used as the mean center temperature of the second image class center, mean_mean2, and mean index of the class center, mean_index2, respectively, to obtain the second image class center C2 = {mean_mean2, mean_index2}.

[0065] After initializing the first and second image cluster centers, class labels are assigned to each original image in the original image set based on the set clustering metric. The clustering metric includes the mean temperature of the cluster centers and the mean index of the cluster centers. After obtaining the initialized first and second image cluster centers, a new one-dimensional label array Label[N] of dimension N is created. The class labels of the original images with image numbers in the first half of the sequence (e.g., original images with image numbers before N / 2, where N / 2 is rounded up) are assigned as the first class labels corresponding to the first image cluster centers, and the class labels of the original images with image numbers in the second half of the sequence (e.g., original images with image numbers after N / 2) are assigned as the second class labels corresponding to the second image cluster centers. Taking a label value of 1 for the first type of label and a label value of 2 for the second type of label as an example, assuming there are 100 original images, and the index values ​​(image numbers) of the original images are arranged from 1 to 100, after initializing the first image class center and the second image class center, the label value of the original images with image numbers 1-50 is set to 1, and the label value of the original images with image numbers 51-100 is set to 2.

[0066] S203: Iteratively update the class labels corresponding to each original image in the original image set based on the clustering metric, and determine the original images corresponding to the first class label and the second class label.

[0067] For example, the mean temperature m and the image index value index of each original image in the statistical information set S1 are traversed. The mean temperature and mean index of the class centers corresponding to the first and second image class centers are used as the clustering metric. The class labels corresponding to each original image in the original image set are updated iteratively. After the iteration, the original images corresponding to the first and second class labels are determined. The original images corresponding to the first and second class labels after the iteration are the original images corresponding to the first and second image class centers after the clustering process.

[0068] Specifically, when iteratively updating the class labels corresponding to each original image in the original image set, the classification can be based on the distance between the average temperature and image index value of the original image and the average temperature and index values ​​of the class centers corresponding to the first and second image class centers. In other words, the original image is assigned to the class label corresponding to the image class center that is closer to the class center. Optionally, the cutoff condition for the iterative update is that the iterative residual between the current iteration result and the previous iteration result is within a set residual threshold range.

[0069] In one embodiment, the class label corresponding to the original image can be iteratively updated based on the weighted distance between the average temperature value and image index value of the original image and the average temperature value and average index value of the class centers corresponding to the class centers of the first and second images. Based on this, such as... Figure 5 The provided schematic diagram illustrates an iterative update process for class labels of original images. This scheme, in iteratively updating the class labels corresponding to each original image in the original image set based on clustering metric, and determining the original images corresponding to the first and second class labels, includes steps S2031-S2032:

[0070] S2031: Based on the first weighted distance of each original image in the original image set relative to the first image class center, and the second weighted distance of each original image in the original image set relative to the second image class center, iteratively update the class label corresponding to each original image in the original image set.

[0071] In each iteration, the original images in the original image set are traversed, and the first weighted distance of each original image to the class center of the first image and the second weighted distance of each original image to the class center of the second image are calculated. For each original image, its corresponding class label is updated to the class label corresponding to the smaller weighted distance. For example, assuming that the current class label of an original image is the first class label, if the first weighted distance calculated in this iteration is less than the second weighted distance, the class label is determined to be the first class label; if the first weighted distance calculated in this iteration is greater than or equal to the second weighted distance, the class label is determined to be the second class label.

[0072] In one embodiment, the first weighted distance is the weighted sum of the difference between the mean temperature and the mean class center temperature of the original image corresponding to the first class label, and the difference between the image index value and the mean class center index of the original image corresponding to the first class label. Correspondingly, the second weighted distance is the weighted sum of the difference between the mean temperature and the mean class center temperature of the original image corresponding to the second class label, and the difference between the image index value and the mean class center index of the original image corresponding to the second class label.

[0073] For example, for any original image I in the original image set i The first weighted distance and the second weighted distance can be calculated using the following formula:

[0074] dist1 = (m i -mean_mean1)+k*(index i -mean_index1)

[0075] dist2=(m i-mean_mean2)+k*(index i -mean_index2)

[0076] Where dist1 is the first weighted distance, dist2 is the second weighted distance, and m i For the original image I i Average temperature, index i For the original image I i The image index value, k is a set floating-point weighted value, which ranges from 0 to 1, for example k = 0.2.

[0077] Furthermore, update the one-dimensional label array Label[N] based on the first weighted distance and the second weighted distance:

[0078]

[0079] S2032: Update the first class center information of the first image class center based on the original image corresponding to the updated first image class center, and update the second class center information of the second image class center based on the original image corresponding to the updated second image class center, until the iteration residual is less than the residual threshold.

[0080] For example, after updating the class labels corresponding to each original image in the original image set in each iteration, the original image corresponding to the updated first image class center updates the first class center information (average class center temperature and average class center index) of the first image class center, and updates the second class center information of the second image class center based on the original image corresponding to the updated second image class center, and repeats steps S2031 and S2032 until the iteration residual is less than the residual threshold. For example, when the difference between the average class center temperature determined in the current iteration and the average class center temperature determined in the previous iteration is within the set average temperature residual threshold, and the difference between the average class center index determined in the current iteration and the average class center index determined in the previous iteration is within the set index average threshold, the original images corresponding to the first and second class labels of the final iteration are determined, and the clustering processing of the original image set is completed.

[0081] S204: Determine the first clustering statistics based on the average temperature of the original image corresponding to the first image cluster center, and determine the second clustering statistics based on the average temperature of the original image corresponding to the second image cluster center.

[0082] In one embodiment, the first clustering statistics include a first cluster mean and a first cluster standard deviation. The first cluster mean is determined based on the mean of the average temperature of the original images corresponding to the first image cluster centers, and the first cluster standard deviation is determined based on the standard deviation of the average temperature of the original images corresponding to the first image cluster centers. Correspondingly, the second clustering statistics include a second cluster mean and a second cluster standard deviation. The second cluster mean is determined based on the mean of the average temperature of the original images corresponding to the second image cluster centers, and the second cluster standard deviation is determined based on the standard deviation of the average temperature of the original images corresponding to the second image cluster centers.

[0083] For example, after clustering the original image set, the class labels of each original image are assigned to the first class label corresponding to the first image class center and the second class label corresponding to the second image class center, respectively. Then, first clustering statistics can be determined based on the original images corresponding to the first class labels, and second clustering statistics can be determined based on the original images corresponding to the second class labels.

[0084] For example, the first cluster mean in the first cluster statistics can be determined based on the following formula:

[0085]

[0086] Where M is the number of original images corresponding to the first type of label, and Label[i] = 1 is the original image corresponding to the first type of label.

[0087] The first cluster standard deviation in the first cluster statistics can be determined based on the following formula:

[0088]

[0089] The second cluster mean in the second cluster statistics can be determined based on the following formula:

[0090]

[0091] Where Label[i] = 2 is the original image corresponding to the second type of label.

[0092] The standard deviation of the second cluster in the second cluster statistics can be determined based on the following formula:

[0093]

[0094] S205: Based on the original statistical information, the first clustering statistical information, and the second clustering statistical information, determine whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0095] In one embodiment, stability can be assessed by evaluating the original kurtosis values ​​of the original image sets before and after clustering. Based on this, such as... Figure 6 A schematic diagram of a thermal stability state determination process is provided. This scheme, based on original statistical information, first cluster statistical information, and second cluster statistical information, determines whether all original images in the original image set were captured when the infrared camera was in a thermally stable state. The process includes steps S2051-S2054:

[0096] S2051: Calculate the original kurtosis evaluation value corresponding to the original image set based on the original statistical information.

[0097] S2052: Calculate the first kurtosis evaluation value corresponding to the first image class center based on the first clustering statistical information.

[0098] S2053: Calculate the second kurtosis evaluation value corresponding to the second image class center based on the second clustering statistical information.

[0099] S2054: Based on the comparison results of the original kurtosis evaluation value, the first kurtosis evaluation value, and the second kurtosis evaluation value, determine whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0100] The original kurtosis evaluation value is calculated based on the statistical mean and statistical standard deviation of the original image set. The original kurtosis evaluation value can be determined using the following formula:

[0101]

[0102] Where N is the number of original images in the original image set, and m i SGmean is the mean temperature of the original image with image index i in the original image set, SGstd is the statistical mean, and SGmean is the statistical standard deviation.

[0103] The first kurtosis evaluation value corresponding to the class center of the first image is calculated based on the first cluster mean and the first cluster standard deviation of all original images corresponding to the first class label after clustering. The first kurtosis evaluation value can be determined based on the following formula:

[0104]

[0105] Where M is the number of images in the original image corresponding to the first image class center, DGmean1 is the first cluster mean, and DGstd1 is the first cluster standard deviation;

[0106] The second kurtosis evaluation value corresponding to the class center of the second image is calculated based on the second cluster mean and second cluster standard deviation of all original images corresponding to the second class label after clustering. The second kurtosis evaluation value can be determined based on the following formula:

[0107]

[0108] Where D is the number of original images corresponding to the second image cluster center, and D = NM, DGmean2 is the second cluster mean, and DGstd2 is the second cluster standard deviation.

[0109] After determining the original kurtosis evaluation value, the first kurtosis evaluation value, and the second kurtosis evaluation value, the original kurtosis evaluation value is compared with the first kurtosis evaluation value and the second kurtosis evaluation value respectively. Based on the comparison results, it is determined whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0110] In one embodiment, when determining whether all original images in the original image set were captured in a thermally stable state of the infrared camera based on the comparison results of the original kurtosis evaluation value, the first kurtosis evaluation value, and the second kurtosis evaluation value, the specific steps are as follows: if the first kurtosis evaluation value is greater than the original kurtosis evaluation value and the second kurtosis evaluation value is greater than the original kurtosis evaluation value, it is determined that not all original images in the original image set were captured in a thermally stable state of the infrared camera; if the first kurtosis evaluation value is less than or equal to the original kurtosis evaluation value, or the second kurtosis evaluation value is less than or equal to the original kurtosis evaluation value, it is determined that all original images in the original image set were captured in a thermally stable state of the infrared camera.

[0111] The original kurtosis evaluation value K1 is compared with the first kurtosis evaluation value K21 and the second kurtosis evaluation value K22. When K21>K1 and K22>K12, it is determined that not all original images in the original image set were captured in the thermally stable state of the infrared camera (i.e., there are original images in the original image set captured within the thermally stable time of the infrared camera). Temperature compensation needs to be performed on the original image set before image stitching. When K21≤K1 or K22≤K12, it is determined that all original images in the original image set were captured in the thermally stable state of the infrared camera. Temperature compensation is not required for the original image set, and image stitching can be performed based on the original image set.

[0112] The above describes a process that determines raw statistical information based on the average temperature of each raw image in the original image set. Then, the original image set is clustered based on the first and second image cluster centers. This process identifies the raw images corresponding to the first and second image cluster centers after clustering. First clustering statistics are determined based on the average temperature of the raw images corresponding to the first image cluster centers, and second clustering statistics are determined based on the average temperature of the raw images corresponding to the second image cluster centers. Finally, the raw statistical information, the first clustering statistics, and the second clustering statistics are used to determine whether all raw images in the original image set were captured under thermally stable conditions using an infrared camera. This allows for a rapid and accurate assessment of the image temperature measurement quality of the original image set, enabling a quick determination of whether temperature compensation is needed before stitching the raw images together. Furthermore, the average temperature and average index of the cluster centers are used as clustering metrics to accurately determine whether two types of temperature data clusters exist in the original image set, allowing for more precise iterative updates of the corresponding class labels for the raw images. Based on the kurtosis evaluation value, the distribution of temperature data in the original image set before and after clustering is judged, and the infrared camera is accurately judged to determine whether it has reached a thermally stable state when the original image is started, so as to ensure the display effect of the stitched image.

[0113] Figure 7 A schematic diagram of an image quality assessment device provided in an embodiment of this application is given. (Reference) Figure 7 The image quality judgment device includes an initial statistics module 31, an image clustering module 32, a clustering statistics module 33, and a stability judgment module 34.

[0114] The initial statistics module 31 is used to determine the average temperature of each original image in the original image set, and to determine the original statistical information based on the average temperature; the image clustering module 32 is used to perform clustering processing on the original image set based on the first image cluster center and the second image cluster center, and to determine the original images corresponding to the first image cluster center and the second image cluster center after clustering processing, respectively; the clustering statistics module 33 is used to determine the first clustering statistics based on the average temperature of the original images corresponding to the first image cluster center, and to determine the second clustering statistics based on the average temperature of the original images corresponding to the second image cluster center; the stability judgment module 34 is used to determine whether all the original images in the original image set were captured in the thermally stable state of the infrared camera based on the original statistical information, the first clustering statistics, and the second clustering statistics.

[0115] The above describes a process that determines raw statistical information based on the average temperature of each raw image in the raw image set. Then, the raw image set is clustered based on the first and second image cluster centers to determine the raw images corresponding to the first and second image cluster centers after clustering. First clustering statistical information is determined based on the average temperature of the raw images corresponding to the first image cluster centers, and second clustering statistical information is determined based on the average temperature of the raw images corresponding to the second image cluster centers. Finally, the raw statistical information, the first clustering statistical information, and the second clustering statistical information are used to determine whether all raw images in the raw image set were captured under thermally stable conditions using an infrared camera. This allows for a rapid and accurate assessment of the image temperature measurement quality of the raw image set, enabling a quick determination of whether temperature compensation is needed before stitching the raw images together.

[0116] In one possible embodiment, the image clustering module 32 is specifically used for:

[0117] Initialize the first image class center and the second image class center, and assign class labels to each original image in the original image set based on the set clustering metric. The class labels include a first class label corresponding to the first image class center and a second class label corresponding to the second image class center.

[0118] Based on the clustering metric, the class labels corresponding to each original image in the original image set are iteratively updated to determine the original images corresponding to the first class label and the second class label.

[0119] In one possible embodiment, when the image clustering module 32 iteratively updates the class labels corresponding to each original image in the original image set based on the clustering metric, and determines the original images corresponding to the first class label and the second class label, it specifically includes:

[0120] Based on the first weighted distance of each original image in the original image set relative to the class center of the first image, and the second weighted distance of each original image in the original image set relative to the class center of the second image, the class label corresponding to each original image in the original image set is iteratively updated;

[0121] The first class center information of the first image class center is updated based on the original image corresponding to the updated first image class center, and the second class center information of the second image class center is updated based on the original image corresponding to the updated second image class center, until the iterative residual is less than the residual threshold.

[0122] In one possible embodiment, the clustering metric includes the mean temperature of the cluster centers and the mean index of the cluster centers;

[0123] The first weighted distance is the weighted sum of the difference between the mean temperature of the original image corresponding to the first class label and the mean temperature of the class center, and the difference between the image index value of the original image corresponding to the first class label and the mean index of the class center.

[0124] The second weighted distance is the weighted sum of the difference between the mean temperature of the original image corresponding to the second class label and the mean temperature of the class center, and the difference between the image index value of the original image corresponding to the second class label and the mean index of the class center.

[0125] In one possible embodiment, the stability determination module 34 is specifically used for:

[0126] Calculate the original kurtosis evaluation value corresponding to the original image set based on the original statistical information;

[0127] Calculate the first kurtosis evaluation value corresponding to the first image class center based on the first clustering statistical information;

[0128] Calculate the second kurtosis evaluation value corresponding to the second image class center based on the second clustering statistical information;

[0129] Based on the comparison results of the original kurtosis evaluation value, the first kurtosis evaluation value, and the second kurtosis evaluation value, it is determined whether all the original images in the original image set were captured in the thermally stable state of the infrared camera.

[0130] In one possible embodiment, when the stability determination module 34 determines whether all the original images in the original image set were captured in the thermally stable state of the infrared camera based on the comparison result of the original kurtosis evaluation value, the first kurtosis evaluation value, and the second kurtosis evaluation value, it specifically includes:

[0131] When the first kurtosis evaluation value is greater than the original kurtosis evaluation value, and the second kurtosis evaluation value is greater than the original kurtosis evaluation value, it is determined that not all original images in the original image set were captured in the thermally stable state of the infrared camera.

[0132] When the first kurtosis evaluation value is less than or equal to the original kurtosis evaluation value, or the second kurtosis evaluation value is less than or equal to the original kurtosis evaluation value, it is determined that all original images in the original image set were captured in the thermally stable state of the infrared camera.

[0133] In one possible embodiment, the raw statistical information includes a statistical mean and a statistical standard deviation, wherein the statistical mean is determined based on the mean of the temperature mean of each raw image in the raw image set, and the statistical standard deviation is determined based on the standard deviation of the temperature mean of each raw image in the raw image set.

[0134] The first clustering statistics include the first cluster mean and the first cluster standard deviation. The first cluster mean is determined based on the mean of the temperature mean of the original image corresponding to the first image cluster center, and the first cluster standard deviation is determined based on the standard deviation of the temperature mean of the original image corresponding to the first image cluster center.

[0135] The second clustering statistics include the second cluster mean and the second cluster standard deviation. The second cluster mean is determined based on the mean of the temperature mean of the original image corresponding to the second image cluster center, and the second cluster standard deviation is determined based on the standard deviation of the temperature mean of the original image corresponding to the second image cluster center.

[0136] In one possible embodiment, the original kurtosis evaluation value is determined based on the following formula:

[0137]

[0138] Where N is the number of original images in the original image set, and m i SGmean is the mean temperature of the original image with image index i in the original image set, SGstd is the statistical mean, and SGmean is the statistical standard deviation.

[0139] The first kurtosis evaluation value is determined based on the following formula:

[0140]

[0141] Where M is the number of images in the original image corresponding to the first image class center, DGmean1 is the first cluster mean, and DGstd1 is the first cluster standard deviation;

[0142] The second kurtosis evaluation value is determined based on the following formula:

[0143]

[0144] Where D is the number of original images corresponding to the second image cluster center, DGmean2 is the second cluster mean, and DGstd2 is the second cluster standard deviation.

[0145] It is worth noting that in the above-described embodiments of the image quality judgment device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of the present invention.

[0146] This application also provides an image quality assessment device, which can integrate the image quality assessment apparatus provided in this application. Figure 8 This is a schematic diagram of the structure of an image quality assessment device provided in an embodiment of this application. (Reference) Figure 8 The image quality assessment device includes an input device 43, an output device 44, a memory 42, and one or more processors 41. The memory 42 stores one or more programs. When the one or more programs are executed by the one or more processors 41, the one or more processors 41 implement the image quality assessment method provided in the above embodiments. The input device 43, output device 44, memory 42, and processors 41 can be connected via a bus or other means. Figure 8 Taking the example of a connection between China and Israel via a bus.

[0147] The memory 42, as a computing device readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as program instructions / modules corresponding to the image quality judgment method described in any embodiment of this application (e.g., the initial statistics module 31, image clustering module 32, clustering statistics module 33, and stability judgment module 34 in the image quality judgment device). The memory 42 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device, etc. Furthermore, the memory 42 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 42 may further include memory remotely located relative to the processor 41, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0148] Input device 43 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 44 may include display devices such as a display screen.

[0149] The processor 41 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 42, thereby realizing the above-mentioned image quality judgment method.

[0150] The image quality assessment device, equipment, and computer provided above can be used to execute the image quality assessment method provided in any of the above embodiments, and have corresponding functions and beneficial effects.

[0151] This application embodiment also provides a storage medium for storing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to execute the image quality judgment method provided in the above embodiments. The image quality judgment method includes: determining the average temperature of each original image in an original image set, and determining original statistical information based on the average temperature; performing clustering processing on the original image set based on a first image class center and a second image class center, and determining the original images corresponding to the first image class center and the second image class center after clustering processing, respectively; determining first clustering statistical information based on the average temperature of the original images corresponding to the first image class center, and determining second clustering statistical information based on the average temperature of the original images corresponding to the second image class center; and determining whether all the original images in the original image set were captured in a thermally stable state by an infrared camera based on the original statistical information, the first clustering statistical information, and the second clustering statistical information.

[0152] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROMs, floppy disks, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disks or optical storage); registers or other similar types of memory elements, etc. Storage media may also include other types of memory or combinations thereof. Furthermore, storage media may reside in a first computer system in which a program is executed, or may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media that may reside in different locations (e.g., in different computer systems connected via a network). Storage media may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0153] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the image quality judgment method described above, but can also execute related operations in the image quality judgment method provided in any embodiment of this application.

[0154] The image quality judgment device, equipment, and storage medium provided in the above embodiments can execute the image quality judgment method provided in any embodiment of this application. For technical details not described in detail in the above embodiments, please refer to the image quality judgment method provided in any embodiment of this application.

[0155] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the claims.

Claims

1. An image quality judging method characterized by comprising: The method comprises: determining a temperature mean value of each original image in an original image set, and determining original statistical information based on the temperature mean value; performing clustering processing on the original image set based on a first image class center and a second image class center, and determining original images corresponding to the first image class center and the second image class center after the clustering processing, respectively; determining first clustering statistical information according to the temperature mean value of the original images corresponding to the first image class center, and determining second clustering statistical information according to the temperature mean value of the original images corresponding to the second image class center; determining whether the original images in the original image set are all taken in a thermal stable state of an infrared camera based on the original statistical information, the first clustering statistical information and the second clustering statistical information.

2. The image quality judging method according to claim 1, characterized by The method comprises: initializing the first image class center and the second image class center, and assigning a class label to each original image in the original image set based on a set clustering metric, the class label comprising a first class label corresponding to the first image class center and a second class label corresponding to the second image class center; iteratively updating the class label corresponding to each original image in the original image set based on the clustering metric, and determining the original images corresponding to the first class label and the second class label.

3. The image quality judging method according to claim 2, characterized by The method comprises: iteratively updating the class label corresponding to each original image in the original image set based on a first weighted distance of each original image in the original image set relative to the first image class center and a second weighted distance of each original image in the original image set relative to the second image class center; updating first class center information of the first image class center based on the original images corresponding to the updated first image class center, and updating second class center information of the second image class center based on the original images corresponding to the updated second image class center, until an iteration residual is less than a residual threshold, the iteration residual comprising a difference between a class center temperature mean value determined in a current iteration and a class center temperature mean value determined in a previous iteration, and a difference between a class center index mean value determined in the current iteration and a class center index mean value determined in the previous iteration.

4. The image quality judging method according to claim 3, characterized by The clustering metric comprises a class center temperature mean value and a class center index mean value; the first weighted distance is a weighted sum result of a difference between a temperature mean value of the original image corresponding to the first class label and the class center temperature mean value, and a difference between an image index value of the original image corresponding to the first class label and the class center index mean value. The second weighted distance is a weighted sum of a difference between a temperature mean value of the original image corresponding to the second class label and the class center temperature mean value, and a difference between an image index value of the original image corresponding to the second class label and the class center index mean value.

5. The image quality judging method according to claim 1, characterized by The determining whether the original images in the original image set are captured in the thermal stable state of the infrared camera based on the original statistical information, the first clustering statistical information and the second clustering statistical information comprises: calculating an original kurtosis evaluation value corresponding to the original image set based on the original statistical information; calculating a first kurtosis evaluation value corresponding to the first image class center based on the first clustering statistical information; calculating a second kurtosis evaluation value corresponding to the second image class center based on the second clustering statistical information; determining whether the original images in the original image set are captured in the thermal stable state of the infrared camera based on a comparison result of the original kurtosis evaluation value, the first kurtosis evaluation value and the second kurtosis evaluation value.

6. The image quality judging method according to claim 5, characterized by The determining whether the original images in the original image set are captured in the thermal stable state of the infrared camera based on the comparison result of the original kurtosis evaluation value, the first kurtosis evaluation value and the second kurtosis evaluation value comprises: when the first kurtosis evaluation value is greater than the original kurtosis evaluation value and the second kurtosis evaluation value is greater than the original kurtosis evaluation value, determining that the original images in the original image set are not captured in the thermal stable state of the infrared camera; when the first kurtosis evaluation value is less than or equal to the original kurtosis evaluation value or the second kurtosis evaluation value is less than or equal to the original kurtosis evaluation value, determining that the original images in the original image set are captured in the thermal stable state of the infrared camera.

7. The image quality judging method according to claim 6, characterized by The original statistical information comprises a statistical mean value and a statistical standard deviation, the statistical mean value is determined based on a mean value of temperature mean values of the original images in the original image set, and the statistical standard deviation is determined based on a standard deviation of the temperature mean values of the original images in the original image set; The first clustering statistical information comprises a first clustering mean value and a first clustering standard deviation, the first clustering mean value is determined based on a mean value of temperature mean values of the original images corresponding to the first image class center, and the first clustering standard deviation is determined based on a standard deviation of the temperature mean values of the original images corresponding to the first image class center; The second clustering statistical information comprises a second clustering mean value and a second clustering standard deviation, the second clustering mean value is determined based on a mean value of temperature mean values of the original images corresponding to the second image class center, and the second clustering standard deviation is determined based on a standard deviation of the temperature mean values of the original images corresponding to the second image class center.

8. The image quality judging method according to claim 7, characterized by The original kurtosis evaluation value is determined based on the following formula: wherein, is the number of images in the original image set, is the temperature mean of the original image with image number i in the original image set, is the statistical mean, is the statistical standard deviation; The first kurtosis evaluation value is determined based on the following formula: wherein, is the number of images of the original images corresponding to the first image class center, is the first cluster mean, is the first cluster standard deviation; The second kurtosis evaluation value is determined based on the following formula: wherein, is the number of images of the original images corresponding to the second cluster center, is the second cluster mean, is the second cluster standard deviation.

9. An image quality judging apparatus characterized by comprising: The method comprises an initial statistical module, an image clustering module, a clustering statistical module and a stability judgment module, wherein: The initial statistical module is configured to determine temperature mean values of the original images in the original image set, and determine original statistical information based on the temperature mean values. The image clustering module is configured to perform clustering processing on the original image set based on the first image class center and the second image class center, and determine original images corresponding to the first image class center and the second image class center after the clustering processing, respectively. The clustering statistics module is configured to determine first clustering statistics information according to temperature means of the original images corresponding to the first image class center, and determine second clustering statistics information according to temperature means of the original images corresponding to the second image class center. The stability determination module is configured to determine whether the original images in the original image set are captured in a thermal stability state of the infrared camera based on the original statistics information, the first clustering statistics information, and the second clustering statistics information.

10. An image quality judging apparatus characterized by comprising: Comprise: a memory and one or more processors; the memory is configured to store one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the image quality determination method according to any one of claims 1-8.

11. A storage medium storing computer-executable instructions, wherein: The computer executable instructions, when executed by a computer processor, are configured to perform the image quality determination method according to any one of claims 1-8. The computer executable instructions, when executed by a computer processor, are configured to perform the image quality determination method according to any one of claims 1-8.

Images