Face on-duty monitoring terminal based on temperature sensing analysis

Abstract

The invention relates to the technical field of image enhancement, in particular to a face on-duty monitoring terminal based on temperature sensing analysis, which comprises an image acquisition module used for acquiring an infrared image in a monitoring area to obtain a grayscale image corresponding to the infrared image; the image partitioning module is used for partitioning the grayscale image into a plurality of small blocks; the block analysis module is used for acquiring internal pixel point distribution characteristic values and neighborhood block distribution characteristic values of the small blocks; the threshold value correction module is used for obtaining an adaptive enhancement degree according to the internal pixel point distribution characteristic values and the neighborhood block distribution characteristic values of the small blocks, and obtaining an adaptive color contrast threshold value according to the adaptive enhancement degree of the small blocks; and the image monitoring module is used for obtaining an enhanced image according to the self-adaptive color contrast threshold value of each small block, performing face on-duty monitoring according to the enhanced image, and improving the image enhancement effect by self-adaptively correcting the clipLimit parameter of each small block.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and in particular to an on-the-job face monitoring terminal based on temperature sensing analysis. Background Art

[0002] For example, the face presence monitoring terminal, such as the electronic work badge on the table, can determine whether the counter staff is on duty by face recognition. At present, the on-duty / off-duty status of counter staff is detected in real time by infrared, which is in line with the development trend of intelligent office in business halls. The face presence monitoring terminal usually collects infrared temperature image data through infrared cameras based on infrared temperature sensing acquisition technology to monitor and judge the real-time status of counter staff. For the collected infrared images, it is usually necessary to use an enhancement algorithm to make the counter staff part of the image clearer in order to obtain more accurate infrared monitoring results. Therefore, it is very important to use image enhancement algorithms to improve the accuracy of the details of the character monitoring results in infrared images.

[0003] The existing limited contrast adaptive histogram equalization algorithm is used to enhance the infrared images collected by the on-the-job face monitoring terminal. The algorithm divides the infrared image into blocks and uses a fixed clipLimit parameter to perform histogram equalization on each block separately. Among them, the clipLimit parameter (color contrast threshold) is used in the algorithm to control the limit of the pixel value in each block to avoid over-enhancement and noise amplification. A smaller clipLimit value reduces the enhancement and produces an overly conservative equalization effect, while a too large clipLimit value may lead to over-enhancement. Therefore, a fixed clipLimit parameter will lead to poor image enhancement effect of infrared images. How to select a suitable clipLimit parameter to improve the image enhancement effect of infrared images has become an urgent problem to be solved. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a face on-the-job monitoring terminal based on temperature sensing analysis to solve the problem of how to select a suitable clipLimit parameter to improve the image enhancement effect of infrared images.

[0005] An embodiment of the present invention provides a face on-the-job monitoring terminal based on temperature sensing analysis, the terminal comprising: An image acquisition module is used to acquire an infrared image in the monitoring area and obtain a grayscale image corresponding to the infrared image; An image blocking module, used to construct a grayscale histogram of the grayscale image, obtain a block size for blocking the grayscale image according to the grayscale histogram, and divide the grayscale image into a plurality of small blocks according to the block size; A block analysis module is used to obtain, for any small block, the internal pixel distribution feature value of the small block according to the gray value difference of each pixel point in the neighborhood range of the small block, and obtain the neighborhood block distribution feature value of the small block according to the difference of the internal pixel distribution feature value and the gray value distribution difference between the small block and the neighboring small block; A threshold correction module is used to obtain the adaptive enhancement degree of the small block according to the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value, and to correct the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain an adaptive color contrast threshold; The image monitoring module is used to obtain the adaptive color contrast threshold of each of the small blocks, perform contrast-limited adaptive histogram equalization on the grayscale image according to the adaptive color contrast threshold of each of the small blocks, obtain an enhanced image, and perform on-the-job face monitoring on the monitoring area according to the enhanced image.

[0006] Preferably, the image segmentation module obtains the segmentation size for segmenting the grayscale image according to the grayscale histogram, including: According to the histogram distribution in the grayscale histogram, obtaining the proportion of the human area in the grayscale image; The difference between the preset maximum block size and the minimum block size is obtained, the multiplication value of the difference and the proportion of the character area is obtained, the sum of the multiplication value and the minimum block size is rounded, and the block size for dividing the grayscale image into blocks is obtained.

[0007] Preferably, the image segmentation module obtains the proportion of the human area in the grayscale image according to the histogram distribution in the grayscale histogram, including: Taking pixels in the grayscale image whose grayscale values ​​are greater than a preset grayscale value threshold as highlight pixels, and obtaining the ratio of the highlight pixels in the grayscale image; In the grayscale histogram, the histogram corresponding to the grayscale value greater than the preset grayscale value threshold is taken as the target histogram, the length of the horizontal coordinate interval corresponding to all the target histograms is obtained, the vertical coordinate value corresponding to each of the target histograms is counted, the target histogram with the vertical coordinate value greater than the preset vertical coordinate threshold is taken as the high histogram, the maximum absolute value of the vertical coordinate value difference between each high histogram and the adjacent histogram is obtained, and the maximum absolute value mean is obtained; The product of the quantity proportion, the length of the horizontal axis interval and the inverse of the maximum absolute value mean is obtained, and the product is normalized to obtain the proportion of the human figure area in the grayscale image.

[0008] Preferably, the block analysis module obtains the internal pixel distribution feature value of the small block according to the gray value difference within the neighborhood range of each pixel in the small block, including: For any pixel point in the small block, the absolute value of the grayscale value difference between each pixel point in the neighborhood range of the pixel point and the pixel point is obtained to obtain the neighborhood grayscale difference degree of the pixel point, and according to the neighborhood grayscale difference degree of each pixel point in the small block, the overall grayscale distribution characteristic value of the small block is obtained; The overall grayscale distribution characteristic value of the small block is normalized to obtain a corresponding normalized value, and the normalized value is used as an independent variable of a preset polynomial function to obtain the internal pixel distribution characteristic value of the small block.

[0009] Preferably, the block analysis module obtains the neighborhood block distribution characteristic value of the small block according to the difference in internal pixel point distribution characteristic value and grayscale distribution between the small block and the neighborhood small block, including: The absolute values ​​of the differences between the internal pixel distribution feature values ​​of the small block and each other small block are obtained in the neighborhood of the small block to obtain the minimum absolute value of the difference, and the ratio between the internal pixel distribution feature value of the small block and the minimum absolute value of the difference is calculated. The absolute values ​​of the grayscale mean differences between the small block and each other small block in the neighborhood of the small block are obtained respectively, and the minimum absolute value of the difference is obtained as the target absolute value of the difference, and the product between the target absolute value of the difference and the ratio is calculated, and the product is normalized to obtain the neighborhood block distribution characteristic value of the small block.

[0010] Preferably, the threshold correction module obtains the adaptive enhancement degree of the small block according to the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value, including: The average of the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value is calculated as the adaptive enhancement degree of the small block.

[0011] Preferably, the threshold correction module corrects the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain the adaptive color contrast threshold, including: The product of the adaptive enhancement degree of the small block and a preset color contrast threshold is obtained as the adaptive color contrast threshold of the small block.

[0012] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention provides a face monitoring terminal based on temperature sensing analysis, comprising an image acquisition module, which is used to acquire an infrared image in a monitoring area and obtain a grayscale image corresponding to the infrared image; an image blocking module, which is used to construct a grayscale histogram of the grayscale image, obtain a block size for blocking the grayscale image according to the grayscale histogram, and divide the grayscale image into a plurality of small blocks according to the block size; a block analysis module, which is used to obtain, for any small block, a distribution feature value of internal pixel points of the small block according to a difference in grayscale values ​​within a neighborhood range of each pixel point in the small block, and obtain a distribution feature value of internal pixel points of the small block according to a difference in internal pixel point distribution feature values ​​between the small block and a neighboring small block; and grayscale distribution difference, obtain the neighborhood block distribution characteristic value of the small block; threshold correction module, used to obtain the adaptive enhancement degree of the small block according to the internal pixel point distribution characteristic value of the small block and the neighborhood block distribution characteristic value, and correct the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain the adaptive color contrast threshold; image monitoring module, used to obtain the adaptive color contrast threshold of each small block, and according to the adaptive color contrast threshold of each small block, limit the contrast adaptive histogram equalization of the gray image to obtain the enhanced image, and perform face on-the-job monitoring on the monitoring area according to the enhanced image. Among them, after the gray image is divided into blocks, the clipLimit parameter of each small block is adaptively corrected, and the adaptive histogram equalization is performed on each small block to obtain the corresponding enhanced image, which increases the clipLimit limit of the block corresponding to the background area and reduces the clipLimit limit of the block corresponding to the character area, thereby enhancing the enhancement effect of the character area, so that the effect of face on-the-job monitoring is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0014] Figure 1 This is a structural block diagram of a face on-the-job monitoring terminal based on temperature sensing analysis provided by the first embodiment of the present invention; Figure 2 It is a schematic diagram of a function curve with an independent variable F and a dependent variable W provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0016] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0017] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0018] The specific scenario targeted by the present invention is: when using a limited contrast histogram equalization algorithm to perform image enhancement on infrared images for on-the-job monitoring, since the human area is mainly monitored, when a unified and fixed clipLimit parameter is used for all areas in the image, the enhancement effect of the human area is poor, and it is possible that the high-temperature area of ​​the background area will be enhanced, thereby interfering with on-the-job monitoring. Therefore, the present invention needs to obtain a better clipLimit parameter for each area for image enhancement.

[0019] See also Figure 1 , is a structural block diagram of a face on-the-job monitoring terminal based on temperature sensing analysis provided by the first embodiment of the present invention, such as Figure 1 As shown, the terminal may include: The image acquisition module 11 is used to acquire an infrared image in the monitoring area and obtain a grayscale image corresponding to the infrared image.

[0020] Use the infrared sensor installed inside the terminal to capture infrared temperature images of employees in the work area in real time. The sampling frequency is 1 second, which is not limited here. In order to improve the effect of face monitoring on the job, after obtaining the infrared image, the infrared image is first grayed to obtain the corresponding gray image, and then the gray image is enhanced using the limited contrast histogram equalization algorithm (CLAHE) to make the human part in the image clearer.

[0021] It should be noted that the CLAHE algorithm achieves image enhancement by dividing the image into multiple tiles and performing histogram equalization on each tile. Unlike the traditional histogram equalization method, the CLAHE algorithm introduces a parameter called "clipLimit", which is the color contrast threshold, to control the limit range of pixel values ​​in each tile. The meaning of clipLimit is the maximum limit allowed for stretching the pixel values ​​in each tile. If the pixel value range of a tile exceeds the range specified by clipLimit, the pixel values ​​in the tile will be evenly stretched to the specified range.

[0022] It can be seen that the size of clipLimit affects the degree of contrast enhancement. A smaller clipLimit value reduces enhancement, while a larger clipLimit value increases enhancement. In the traditional CLAHE algorithm, each small block in the grayscale image is enhanced by a fixed clipLimit value, resulting in poor image enhancement effect of the infrared image. Therefore, in the embodiment of the present invention, the clipLimit value of each small block in the grayscale image is adaptively obtained to achieve the best enhancement effect. The specific enhancement process is described below.

[0023] The image blocking module 12 is used to construct a grayscale histogram of the grayscale image, obtain a block size for blocking the grayscale image according to the grayscale histogram, and divide the grayscale image into multiple small blocks according to the block size.

[0024] Before adaptively obtaining the clipLimit value of each small block in the grayscale image, the grayscale image needs to be divided into blocks. The size of the blocks determines the enhancement effect of the subsequent histogram equalization within each block. Since the contrast-limited histogram equalization in the present invention is mainly to enhance the character area, the size of the blocks depends on the possible proportion of the character area in the grayscale image. When the character is at a farther position in the grayscale image, the character area is small, and a relatively small block should be used. On the contrary, when the character is at a closer position in the grayscale image, the character area is large, and a relatively large block should be used, thereby ensuring that the subsequent histogram equalization within the block has the effect of enhancing the character area.

[0025] In order to optimally block the grayscale image, first construct a grayscale histogram of the grayscale image, where the horizontal axis of the grayscale histogram is the grayscale value from 0 to 255 with a unit interval of 1, and the vertical axis is the number of pixels corresponding to the grayscale value. Then, the block size for blocking the grayscale image is obtained according to the grayscale histogram, which is also the optimal block size. Then, the grayscale image is divided into multiple small blocks according to the block size, so that the blocking of the grayscale image is more accurate.

[0026] The specific process of obtaining the block size for dividing the grayscale image into blocks according to the grayscale histogram is as follows: (1) According to the histogram distribution in the grayscale histogram, the proportion of the human area in the grayscale image is obtained.

[0027] Specifically, pixels in the grayscale image whose grayscale values ​​are greater than a preset grayscale value threshold are taken as highlight pixels, and the ratio of the highlight pixels in the grayscale image is obtained; In the grayscale histogram, the histogram corresponding to the grayscale value greater than the preset grayscale value threshold is taken as the target histogram, the length of the horizontal coordinate interval corresponding to all the target histograms is obtained, the vertical coordinate value corresponding to each of the target histograms is counted, the target histogram with the vertical coordinate value greater than the preset vertical coordinate threshold is taken as the high histogram, the maximum absolute value of the vertical coordinate value difference between each high histogram and the adjacent histogram is obtained, and the maximum absolute value mean is obtained; The product of the quantity proportion, the length of the horizontal axis interval and the inverse of the maximum absolute value mean is obtained, and the product is normalized to obtain the proportion of the human figure area in the grayscale image.

[0028] In one embodiment, the gray value threshold is preset to 100, and the pixels with gray values ​​greater than 100 in the gray image are taken as highlight pixels, and the number ratio of highlight pixels in the gray image is obtained. ,in, represents the number of highlighted pixels in the grayscale image, and N represents the total number of pixels in the grayscale image. The histogram formed by the highlighted pixels in the grayscale histogram is taken as the target histogram, and the distribution range of the target histogram in the grayscale histogram is recorded as , which is the length of the horizontal axis interval corresponding to the horizontal axis in the grayscale histogram.

[0029] The number of pixels corresponding to the target histogram (that is, the ordinate value) greater than 20 is taken as a high histogram. For any i-th high histogram, the absolute value of the difference between the ordinate values ​​of the i-th high histogram and its two adjacent histograms on the left and right is obtained in the grayscale histogram to obtain the maximum absolute value. Therefore, according to the proportion of quantity , the distribution range of the target histogram in the grayscale histogram The maximum absolute value corresponding to each high bar , get the proportion of the character area in the grayscale image:

[0030] Among them, R represents the proportion of the human area in the grayscale image, norm() represents the normalization function, Represents the number of highlighted pixels in the grayscale image, N represents the total number of pixels in the grayscale image, Indicates the length of the horizontal axis interval, represents the number of tall rectangular columns, Represents the maximum absolute value corresponding to the i-th high bar.

[0031] It should be noted that the human region, as a heat source, is usually a highlight region in the infrared image. Therefore, the more pixels in the highlight region, the larger the human region may be. The larger the value of , the greater the value of . However, there may be other heat sources in the infrared image, such as computers and other plug-in office equipment. Therefore, the possible size of the human area cannot be judged solely by the proportion of the highlight area. The surface temperature of the equipment is concentrated and stabilized at a fixed value, which corresponds to the grayscale histogram as several high-height histograms concentrated in the highlight area. The distribution range of the high-height histograms is small, and the difference between the other adjacent histograms is large; compared with computers and other equipment, the temperature range of the human body surface is relatively wider, which is manifested in the grayscale histogram as the distribution of the high-height histograms in the highlight area is wider, and the difference between the other adjacent histograms is relatively mild. Therefore, the more the data proportion of the highlight area in the grayscale histogram, the wider the distribution of the high histograms, and the smaller the difference between the histograms corresponding to each adjacent highlight data, the greater the suspected proportion of the distribution of the human area in the grayscale image, that is, The larger the value, The bigger the The larger the value is, the greater the proportion of the human area in the corresponding grayscale image.

[0032] (2) Obtaining the difference between the preset maximum block size and the minimum block size, obtaining the multiplication value of the difference and the proportion of the character area, rounding the sum of the multiplication value and the minimum block size, and obtaining the block size for dividing the grayscale image.

[0033] In one embodiment, the calculation expression of the block size for dividing the grayscale image into blocks is:

[0034] Among them, M represents the block size, R represents the proportion of the human area in the grayscale image, max represents the preset maximum block size, and min represents the preset minimum block size. Indicates the round-up symbol.

[0035] It should be noted that the classic value range of the block size is between 8×8 and 128×128 pixels, corresponding to the setting of max to 128 and min to 8; the larger the proportion of the human area, the larger the area of ​​the human figure in the grayscale image, and the corresponding larger block size is collected. That is, the larger the proportion of the human area, the larger the block size, and the better the histogram equalization effect of the subsequent small blocks.

[0036] At this point, through the highlight distribution characteristics of the high-temperature human area in the infrared image, combined with the difference in grayscale distribution characteristics from the non-human high-temperature equipment, a more accurate degree of suspected proportion of the human area is obtained, so that the human area can be divided into blocks more accurately according to the suspected proportion, thereby improving the enhancement effect of the subsequent histogram equalization within the block on the human area.

[0037] The block analysis module 13 is used to obtain the internal pixel distribution characteristic value of any small block according to the grayscale value difference within the neighborhood range of each pixel point in the small block, and obtain the neighborhood block distribution characteristic value of the small block according to the difference in internal pixel distribution characteristic value and grayscale distribution between the small block and the neighboring small blocks.

[0038] After the grayscale image is divided into multiple small blocks, the clipLimit value is used to limit the contrast of each small block, thereby performing histogram equalization. However, the clipLimithi value of each small block is different, and the corresponding enhancement degree is different. Therefore, in the embodiment of the present invention, the clipLimit value of each small block is adaptively corrected to obtain an adaptive clipLimit value (adaptive color contrast threshold) to increase the clipLimit limit of the background part blocks and reduce the clipLimit limit of the main character part blocks.

[0039] Taking a small block as an example, the method for obtaining the adaptive color contrast threshold of the small block is: (1) Obtaining internal pixel distribution feature values ​​of the small block according to the grayscale value difference within the neighborhood range of each pixel in the small block.

[0040] Specifically, for any pixel point in the small block, the absolute value of the grayscale value difference between each pixel point in the neighborhood range of the pixel point and the pixel point is obtained to obtain the neighborhood grayscale difference degree of the pixel point, and according to the neighborhood grayscale difference degree of each pixel point in the small block, the overall grayscale distribution characteristic value of the small block is obtained; The overall grayscale distribution characteristic value of the small block is normalized to obtain a corresponding normalized value, and the normalized value is used as an independent variable of a preset polynomial function to obtain the internal pixel distribution characteristic value of the small block.

[0041] In one embodiment, taking the pth small block as an example, the absolute value of the grayscale value difference between the yth pixel point in the pth small block and the xth pixel point in the eight neighborhoods of the yth pixel point is calculated, which is recorded as , according to the absolute value of the gray value difference between each pixel in the p-th small block and each pixel in its eight neighborhoods, the overall gray distribution characteristic value of the p-th small block is obtained:

[0042] in, represents the overall grayscale distribution characteristic value of the pth small block, norm() represents the normalization function, Indicates the number of pixels in the pth small block, Represents the number of pixels in the eight-neighborhood of the y-th pixel, It represents the absolute value of the grayscale difference between the yth pixel in the pth small block and the xth pixel in the eight-neighborhood of the yth pixel.

[0043] It should be noted that heat source devices such as computers have stable surface heat dissipation and maintain a fixed heat dissipation degree. When mapped in the infrared temperature image, the grayscale values ​​of multiple pixels in a smaller block area are more similar. Compared with the temperature of the background area, the grayscale values ​​of multiple pixels on the surface of the human body are more similar. However, the heat dissipation of the human body surface is uneven compared with that of the equipment, and there is a certain slight fluctuation. When mapped in the infrared temperature image, the grayscale values ​​of multiple pixels in a smaller block area are more similar, and the grayscale values ​​are less similar. Therefore, using Characterize the similarity between the y-th pixel and its neighboring pixels. Correspondingly, use A comprehensive index that characterizes the neighborhood similarity of all pixels in the pth small block. The closer the value of is to the minimum, the more the p-th small block conforms to the distribution characteristics of the same pixels within the block of the high-temperature equipment area. The closer the value is to the maximum, the more the p-th small block conforms to the distribution characteristics of the background area with many chaotic noise points and few similar pixels within the block.

[0044] In order to make the overall grayscale distribution feature value of the pth small block more accurately represent the pixel distribution characteristics of the pth small block belonging to the character area, therefore, the following is constructed: Figure 2 The function shown is the function with independent variable F and dependent variable W ,Depend on Figure 2 It can be seen that the value range of the function curve is in the interval [0, 1], respectively and The minimum value is reached at , thereby excluding the internal distribution characteristics of the blocks with background noise and high-temperature equipment, and the internal pixel distribution characteristic value is larger, and the distribution of the p-th small block is more consistent with the distribution of the human area. Therefore, the overall grayscale distribution characteristic value of the p-th small block is input into the function In the above example, we can get the corresponding internal pixel distribution characteristic value .

[0045] (2) Obtaining the neighborhood block distribution characteristic value of the small block according to the difference in internal pixel distribution characteristic value and grayscale distribution between the small block and the neighborhood small block.

[0046] Specifically, the absolute values ​​of the differences of the internal pixel distribution feature values ​​between the small block and each other small block are obtained in the neighborhood of the small block, the minimum absolute value of the difference is obtained, and the ratio between the internal pixel distribution feature value of the small block and the minimum absolute value of the difference is calculated. The absolute values ​​of the grayscale mean differences between the small block and each other small block in the neighborhood of the small block are obtained respectively, and the minimum absolute value of the difference is obtained as the target absolute value of the difference, and the product between the target absolute value of the difference and the ratio is calculated, and the product is normalized to obtain the neighborhood block distribution characteristic value of the small block.

[0047] In one embodiment, since the temperature of different parts of the human body surface is quite different, while the temperature of different parts of the equipment surface is basically similar, there is a high probability that there are blocks with large grayscale differences around the blocks in the human body area, while other blocks around the high-temperature equipment area block have a high grayscale similarity. For the pth small block, in the eight neighborhoods of the pth small block, the absolute value of the difference between the internal pixel distribution feature values ​​of the pth small block and each other small block (except the pth small block) is obtained to obtain the minimum absolute value of the difference. At the same time, according to the grayscale values ​​of all pixels in each small block, the grayscale mean of each small block is obtained respectively. In the eight neighborhoods of the pth small block, the absolute value of the difference between the grayscale mean of the pth small block and each other small block (except the pth small block) is obtained respectively, and the minimum absolute value of the difference is obtained. , according to the minimum absolute value of the difference and the minimum absolute value of the difference , get the neighborhood block distribution characteristic value of the pth small block:

[0048] in, represents the neighborhood block distribution characteristic value of the pth small block, Represents the internal pixel distribution feature value of the pth small block, Indicates the internal pixel distribution characteristic values ​​of other small blocks corresponding to the minimum absolute value of the difference. Indicates the grayscale mean of other small blocks corresponding to the minimum absolute value of the difference, It represents the grayscale mean of the pth small block, and norm() represents the normalization function.

[0049] It should be noted that The larger the value of is, the greater the probability that the pth small block belongs to the character area, corresponding to The smaller it is, the more the p-th block and the neighboring blocks belong to the same type of blocks, the stronger the distribution characteristics of the neighboring blocks corresponding to the p-th block are, and the more likely it is a character area block; The larger the value of , the greater the grayscale difference between the blocks around the p-th small block, the more consistent with the distribution characteristics of the block neighborhood of the character area, and the larger the distribution characteristic value of the neighborhood block corresponding to the p-th small block.

[0050] The threshold correction module 14 is used to obtain the adaptive enhancement degree of the small block according to the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value, and to correct the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain an adaptive color contrast threshold.

[0051] After obtaining the internal pixel distribution characteristic value and the neighborhood block distribution characteristic value of the p-th small block, considering that the larger the internal pixel distribution characteristic value, the larger the neighborhood block distribution characteristic value, the more the P-th small block belongs to the block of the character area, and then according to the internal pixel distribution characteristic value and the neighborhood block distribution characteristic value, the adaptive enhancement degree of the P-th small block is obtained: the mean of the internal pixel distribution characteristic value and the neighborhood block distribution characteristic value of the small block is calculated as the adaptive enhancement degree of the small block.

[0052] Among them, the degree of adaptive enhancement The calculation expression is:

[0053] After determining the adaptive enhancement degree of the pth small block, the adaptive color contrast threshold of the pth small block is obtained, specifically: the product of the adaptive enhancement degree of the small block and the preset color contrast threshold is obtained as the adaptive color contrast threshold of the small block.

[0054] Among them, the calculation expression of the adaptive color contrast threshold is:

[0055] in, Represents the adaptive color contrast threshold of the pth small block, which is also the corrected clipLimit value. Indicates the fixed clipLimit value in the traditional CLAHE algorithm, that is, the clipLimit value before correction. Indicates the degree of adaptive enhancement of the p-th small block.

[0056] It should be noted that the greater the degree of adaptive enhancement, the more the corresponding small block belongs to the human area, the more the clipLimit value needs to be increased, which can reduce the contrast limit of the blocks in the human area. Conversely, the smaller the degree of adaptive enhancement, the less the corresponding small block belongs to the human area, the more the clipLimit value needs to be reduced, which can increase the contrast limit of the blocks in the human area.

[0057] Similarly, according to the method for obtaining the adaptive color contrast threshold of the p-th small block, the adaptive color contrast threshold of each small block in the grayscale image is obtained respectively.

[0058] The image monitoring module 15 is used to obtain the adaptive color contrast threshold of each of the small blocks, perform contrast-limited adaptive histogram equalization on the grayscale image according to the adaptive color contrast threshold of each of the small blocks, obtain an enhanced image, and perform on-the-job face monitoring on the monitoring area according to the enhanced image.

[0059] After the adaptive color contrast threshold of each small block is obtained as described above, histogram equalization can be performed on each small block according to the adaptive color contrast threshold of each small block, so as to obtain an enhanced image corresponding to the grayscale image after image enhancement. At this time, the enhanced image excludes the high temperature area corresponding to the background area and other non-human areas, and enhances the human area in the image. Among them, performing histogram equalization on each small block according to the adaptive color contrast threshold belongs to the prior art in the CLAHE algorithm, which will not be repeated here.

[0060] Since the characters in the enhanced image are more clearly reflected, in the enhanced image, on-the-job face monitoring is performed on the monitoring area according to the enhanced image, so as to achieve more accurate on-the-job monitoring analysis results. It is worth noting that on-the-job face monitoring in the monitoring area according to the enhanced image belongs to the existing technology and will not be elaborated here.

[0061] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A face monitoring terminal based on temperature sensing analysis, characterized in that: The terminal comprises: An image acquisition module is used to acquire an infrared image in the monitoring area and obtain a grayscale image corresponding to the infrared image; An image blocking module, used to construct a grayscale histogram of the grayscale image, obtain a block size for blocking the grayscale image according to the grayscale histogram, and divide the grayscale image into a plurality of small blocks according to the block size; A block analysis module is used to obtain, for any small block, the internal pixel distribution feature value of the small block according to the gray value difference of each pixel point in the neighborhood range of the small block, and obtain the neighborhood block distribution feature value of the small block according to the difference of the internal pixel distribution feature value and the gray value distribution difference between the small block and the neighboring small block; A threshold correction module is used to obtain the adaptive enhancement degree of the small block according to the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value, and to correct the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain an adaptive color contrast threshold; The image monitoring module is used to obtain the adaptive color contrast threshold of each of the small blocks, perform contrast-limited adaptive histogram equalization on the grayscale image according to the adaptive color contrast threshold of each of the small blocks, obtain an enhanced image, and perform on-the-job face monitoring on the monitoring area according to the enhanced image.

2. The face on-the-job monitoring terminal based on temperature sensing analysis according to claim 1 is characterized in that: The image segmentation module obtains the segmentation size for segmenting the grayscale image according to the grayscale histogram, including: According to the histogram distribution in the grayscale histogram, obtaining the proportion of the human area in the grayscale image; The difference between the preset maximum block size and the minimum block size is obtained, the multiplication value of the difference and the proportion of the character area is obtained, the sum of the multiplication value and the minimum block size is rounded, and the block size for dividing the grayscale image into blocks is obtained.

3. The face monitoring terminal based on temperature sensing analysis according to claim 2 is characterized in that: The image segmentation module obtains the proportion of the human area in the grayscale image according to the histogram distribution in the grayscale histogram, including: Taking pixels in the grayscale image whose grayscale values ​​are greater than a preset grayscale value threshold as highlight pixels, and obtaining the ratio of the highlight pixels in the grayscale image; In the grayscale histogram, the histogram corresponding to the grayscale value greater than the preset grayscale value threshold is taken as the target histogram, the length of the horizontal coordinate interval corresponding to all the target histograms is obtained, the vertical coordinate value corresponding to each of the target histograms is counted, the target histogram with the vertical coordinate value greater than the preset vertical coordinate threshold is taken as the high histogram, the maximum absolute value of the vertical coordinate value difference between each high histogram and the adjacent histogram is obtained, and the maximum absolute value mean is obtained; The product of the quantity proportion, the length of the horizontal axis interval and the inverse of the maximum absolute value mean is obtained, and the product is normalized to obtain the proportion of the human figure area in the grayscale image.

4. The face monitoring terminal based on temperature sensing analysis according to claim 1 is characterized in that: The block analysis module obtains the internal pixel distribution feature value of the small block according to the gray value difference within the neighborhood range of each pixel point in the small block, including: For any pixel point in the small block, the absolute value of the grayscale value difference between each pixel point in the neighborhood range of the pixel point and the pixel point is obtained to obtain the neighborhood grayscale difference degree of the pixel point, and according to the neighborhood grayscale difference degree of each pixel point in the small block, the overall grayscale distribution characteristic value of the small block is obtained; The overall grayscale distribution characteristic value of the small block is normalized to obtain a corresponding normalized value, and the normalized value is used as an independent variable of a preset polynomial function to obtain the internal pixel distribution characteristic value of the small block.

5. The face on-the-job monitoring terminal based on temperature sensing analysis according to claim 1 is characterized in that: The block analysis module obtains the neighborhood block distribution characteristic value of the small block according to the difference in internal pixel distribution characteristic value and grayscale distribution between the small block and the neighborhood small block, including: The absolute values ​​of the differences between the internal pixel distribution feature values ​​of the small block and each other small block are obtained in the neighborhood of the small block to obtain the minimum absolute value of the difference, and the ratio between the internal pixel distribution feature value of the small block and the minimum absolute value of the difference is calculated. The absolute values ​​of the grayscale mean differences between the small block and each other small block in the neighborhood of the small block are obtained respectively, and the minimum absolute value of the difference is obtained as the target absolute value of the difference, and the product between the target absolute value of the difference and the ratio is calculated, and the product is normalized to obtain the neighborhood block distribution characteristic value of the small block.

6. The face monitoring terminal based on temperature sensing analysis according to claim 1 is characterized in that: The threshold correction module obtains the adaptive enhancement degree of the small block according to the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value, including: The average of the internal pixel distribution characteristic value of the small block and the neighborhood block distribution characteristic value is calculated as the adaptive enhancement degree of the small block.

7. The face monitoring terminal based on temperature sensing analysis according to claim 1 is characterized in that: The threshold correction module corrects the preset color contrast threshold according to the adaptive enhancement degree of the small block to obtain an adaptive color contrast threshold, including: The product of the adaptive enhancement degree of the small block and a preset color contrast threshold is obtained as the adaptive color contrast threshold of the small block.

Images