Security inspection monitoring image enhancement method and system based on machine vision

Abstract

The invention relates to the technical field of image processing, in particular to a security inspection monitoring image enhancement method and system based on machine vision. The method comprises the following steps: collecting each frame of security check monitoring image, and obtaining a dynamic area of each frame of security check monitoring image; obtaining the behavior movement complexity of the dynamic region according to the area change and the angular point number of the dynamic region; performing clustering analysis on the angular points of the dynamic region to obtain behavior confusion of the dynamic region; based on the behavior motion complexity and the behavior confusion, acquiring the behavior complexity of the dynamic region; obtaining the fuzzy degree of a dynamic region according to the behavior complexity; and acquiring an enhancement coefficient of the current frame of security check monitoring image according to the fuzzy degrees of all the dynamic regions of the current frame of security check monitoring image, and enhancing the acquired current frame of security check monitoring image according to the enhancement coefficient to obtain a current frame of security check monitoring enhanced image. According to the invention, the action behavior of the person can be identified more easily after the image is enhanced.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for enhancing security inspection and monitoring images based on machine vision. Background Technology

[0002] With rapid societal development and continuous technological advancements, the application of security monitoring systems in public safety has significantly expanded, particularly in transportation hubs such as airports, train stations, and subway stations. These locations are typically key areas for public safety management and emergency response; therefore, efficient, real-time monitoring image processing and analysis systems have become indispensable tools for ensuring public safety. In these environments, security monitoring systems not only need to handle monitoring data from large crowds but also address potential emergencies or security threats. For example, in train stations or subway stations, monitoring systems need to quickly respond to any abnormal passenger behavior to prevent stampedes or other emergencies. Therefore, to ensure safety in these complex environments, traditional security monitoring systems can no longer meet the growing demands, especially when faced with large amounts of complex real-time image data. Effective real-time image processing and analysis can not only provide detailed dynamic images but also quickly identify abnormal behavior and potential threats.

[0003] In security monitoring systems, due to the frequent and complex activities of people in the monitored area, and the inability of cameras to capture clear images in a short time when people move quickly, motion blur is likely to appear in security monitoring videos. This blurring phenomenon causes the loss of important dynamic information in the security monitoring images, making it difficult to identify the actions and behaviors of people in the monitored area and to accurately extract potential threats. Summary of the Invention

[0004] To address the technical problem that the monitoring area is complex and people move quickly, making it difficult to capture clear images in a short time, resulting in motion blur in the images and making it difficult to recognize people's actions, this invention provides a security inspection monitoring image enhancement method and system based on machine vision.

[0005] In a first aspect, the present invention provides a security inspection and monitoring image enhancement method based on machine vision, which adopts the following technical solution: A machine vision-based image enhancement method for security inspection surveillance includes the following steps: Acquire each frame of security check surveillance image; obtain each dynamic region of each frame of security check surveillance image; obtain the behavioral motion complexity of each dynamic region in the current frame of security check surveillance image based on the area change and the number of corner points of the dynamic region; obtain the feature vector of each corner point of each dynamic region in the current frame of security check surveillance image; cluster the dynamic regions and corner points of each dynamic region in the current frame of security check surveillance image based on the feature vectors to obtain several clusters; obtain the behavioral disorder of each dynamic region in the current frame of security check surveillance image based on the number and distribution of clusters; multiply the behavioral motion complexity and behavioral disorder to obtain the behavioral complexity degree of each dynamic region in the current frame of security check surveillance image. Based on the complexity of the behavior, the blur level of each dynamic region of the current frame security inspection monitoring image is obtained; based on the blur level, the enhancement coefficient of the current frame security inspection monitoring image is obtained, and the current frame security inspection monitoring image is enhanced to obtain the current frame enhanced security inspection monitoring image.

[0006] The innovation of this invention lies in first acquiring each dynamic region of each frame of security monitoring image. Extracting dynamic regions helps to filter out active parts from static images for subsequent analysis. Based on the changes in the area of ​​the dynamic region and the number of corner points, the complexity of behavioral motion is obtained. Based on the number of behavioral pattern types in the dynamic region, the disorder of behavior in the dynamic region is obtained. Then, by comprehensively considering the product of behavioral motion complexity and disorder, the overall behavioral complexity of the dynamic region can be evaluated, and the degree of blurring in the dynamic region can be accurately assessed. Furthermore, based on the degree of blurring of the dynamic region, the enhancement coefficient of the dynamic region is obtained and the image is deblurred to ensure that security personnel can clearly see important details and identify abnormal behavior more quickly and efficiently, and respond to emergencies in a timely manner.

[0007] Preferably, acquiring each dynamic region of each frame of security monitoring image includes: A preset difference threshold T is used. The absolute value of the grayscale difference between any pixel in the current frame of the security monitoring image and the pixel at the same position in the previous frame of the security monitoring image is recorded as the first difference. If the ratio of the first difference to 255 is greater than or equal to the difference threshold T, the pixel in the current frame of the security monitoring image is recorded as a dynamic pixel. The region formed by all adjacent dynamic pixels in the current frame of the security monitoring image is recorded as a dynamic region, thus obtaining each dynamic region of the current frame of the security monitoring image.

[0008] Extracting dynamic regions helps to filter out active parts from static images, reducing interference from irrelevant regions for subsequent analysis.

[0009] Preferably, the step of obtaining the behavioral motion complexity of each dynamic region in the current frame security inspection monitoring image includes: , This represents the behavioral motion complexity of the i-th dynamic region in the current frame of the security inspection monitoring image. This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image; This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image, corresponding to the dynamic region in the previous frame of the security inspection monitoring image. This represents the number of corner points in the i-th dynamic region of the current frame security monitoring image; norm() represents the normalization function.

[0010] By monitoring changes in the area and number of corner points of a dynamic region, the complexity of behavioral motion can be captured, which can help identify ambiguities in the dynamic region.

[0011] Preferably, obtaining the feature vector of each corner point of each dynamic region in the current frame security monitoring image includes: Obtain the matching corner points of each corner point in each dynamic region of the current frame security inspection monitoring image; The Euclidean distance between the position coordinates of the j-th corner point in the i-th dynamic region of the current frame security inspection image and the position coordinates of its matching corner point is denoted as the movement distance of the j-th corner point; the angle between the direction of the line connecting the position coordinates of the j-th corner point in the i-th dynamic region of the current frame security inspection image and its matching corner point and the horizontal direction is denoted as the movement direction of the j-th corner point; the movement distance, movement direction, and coordinate position of the j-th corner point in the i-th dynamic region of the current frame security inspection image are used as the feature vector of the j-th corner point.

[0012] Preferably, the step of obtaining the matching corner points of each corner point in each dynamic region of the current frame security inspection monitoring image includes: For each dynamic region of each frame of security inspection monitoring image, a SIFT descriptor is generated. Then, a feature matching algorithm is used to match the corner points of each dynamic region of the current frame of security inspection monitoring image with the corresponding dynamic region of the previous frame of security inspection monitoring image that have similar SIFT descriptors, so as to obtain the matching corner points of each corner point of each dynamic region of the current frame of security inspection monitoring image.

[0013] Preferably, the step of acquiring the behavioral disorder of each dynamic region of the current frame security inspection monitoring image includes: Obtain the neighboring clusters of each cluster in each dynamic region of the current frame security monitoring image; , The behavioral disorder of the i-th dynamic region in the current frame of the security inspection and surveillance image; The Euclidean distance between the nth cluster of the i-th dynamic region of the current frame security monitoring image and the cluster center of its m-th neighboring cluster; The number of neighboring clusters of the nth cluster representing the i-th dynamic region of the current frame security inspection monitoring image; The number of clusters in the i-th dynamic region of the current frame security monitoring image represents the number of clusters; exp() represents the exponential function with the natural constant as the base; norm() represents the normalization function.

[0014] Behavioral disorder reflects the number of movement types of people within the monitored area. A high level of disorder may indicate the presence of numerous behavioral patterns within the area, making the area more likely to be ambiguous.

[0015] Preferably, obtaining the blur level of each dynamic region in the current frame security monitoring image includes: Obtain the sequence of behavioral complexity of the i-th dynamic region in the current frame of the security inspection monitoring image; , This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The behavior complexity of the i-th dynamic region in the current frame of the security inspection and surveillance image; The number of data points in the sequence representing the behavioral complexity of the i-th dynamic region of the current frame security monitoring image; The (d+1)th data value in the sequence of behavioral complexity of the i-th dynamic region of the current frame security inspection monitoring image; The d-th data value represents the sequence of behavioral complexity of the i-th dynamic region in the current frame of the security inspection monitoring image; norm() represents the normalization function.

[0016] By inferring the degree of image blur from the complexity of the behavior, it is possible to determine whether image clarity needs to be enhanced, effectively avoiding difficulties in behavior recognition caused by unclear images.

[0017] Preferably, the step of obtaining the behavioral complexity sequence of the i-th dynamic region of the current frame security monitoring image includes: The number of historical frame images is preset to G; the matching relationship between the dynamic regions of the current frame security monitoring image and the previous G frame security monitoring images is obtained according to the optical flow method, and the dynamic regions corresponding to each dynamic region of the current frame security monitoring image in the previous G frame security monitoring images are obtained; the behavioral complexity of the i-th dynamic region of the current frame security monitoring image and the previous G frame security monitoring images is arranged according to the image acquisition order, and the behavioral complexity sequence of the i-th dynamic region of the current frame security monitoring image is obtained.

[0018] Preferably, the step of obtaining the enhancement coefficient of the current frame security inspection and monitoring image, and enhancing the current frame security inspection and monitoring image to obtain the current frame enhanced security inspection and monitoring image includes: , This represents the enhancement factor of the current frame of the security surveillance image; This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The mean value representing the blur level of all dynamic regions in the current frame of the security inspection and surveillance image; This represents the number of dynamic regions in the current frame of the security inspection image. The enhancement coefficient of the current frame is used as the sharpening intensity of each pixel within all dynamic regions of the current frame, while the sharpening intensity of each pixel in the non-dynamic regions of the current frame is set to 0. Based on the sharpening intensity of each pixel in the current frame, the Laplacian sharpening algorithm is used to enhance the current frame, resulting in the enhanced security inspection image.

[0019] Secondly, the present invention provides a security inspection and monitoring image enhancement system based on machine vision, which adopts the following technical solution: A machine vision-based security inspection and monitoring image enhancement system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the aforementioned machine vision-based security inspection and monitoring image enhancement method.

[0020] By adopting the above technical solution, a computer program is generated from the above-mentioned machine vision-based security monitoring image enhancement method and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0021] The present invention has the following technical effects: First, the present invention acquires each dynamic region of each frame of security inspection monitoring image, and obtains the behavioral motion complexity based on the change in the area of ​​the dynamic region and the number of corner points; it obtains the behavioral disorder of the dynamic region based on the number of types of personnel movement behavior in the dynamic region; then, it comprehensively considers the product of behavioral motion complexity and disorder, which helps to comprehensively evaluate the behavioral complexity of the dynamic region and accurately assess the blurring degree in the dynamic region; furthermore, based on the blurring degree of the dynamic region, it obtains the enhancement coefficient of the dynamic region and performs deblurring processing on the image, ensuring that security personnel can clearly see important details and identify abnormal behavior more quickly and efficiently, and respond to emergencies in a timely manner. Attached Figure Description

[0022] Figure 1 This is a flowchart of a security inspection and monitoring image enhancement method based on machine vision according to an embodiment of the present invention. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0024] This invention discloses a machine vision-based image enhancement method for security inspection monitoring, referring to... Figure 1 This includes steps S1-S4: S1: Acquire each frame of security inspection monitoring image.

[0025] In this embodiment of the invention, a camera is installed at the security checkpoint to capture each frame of the security check process, which is recorded as each frame of security monitoring image.

[0026] S2: Obtain each dynamic region of each frame of security inspection monitoring image; based on the area change and the number of corner points of the dynamic region, obtain the behavioral motion complexity of each dynamic region in the current frame of security inspection monitoring image; cluster the corner points of the dynamic region to obtain several clusters; based on the number and distribution of clusters, obtain the behavioral disorder of each dynamic region in the current frame of security inspection monitoring image; based on the behavioral motion complexity and behavioral disorder, obtain the behavioral complexity of each dynamic region in the current frame of security inspection monitoring image.

[0027] It should be noted that in the monitored area, due to the complexity of people and their rapid movement, cameras cannot capture clear images in a short time, which easily leads to motion blur in the security monitoring video images, making it difficult to identify the actions of people in the monitored area. Therefore, this invention first identifies the dynamic areas in the security monitoring images, analyzes the movement of people in the dynamic areas, obtains the complexity of the behavior in the dynamic areas, and then evaluates the blur level of the dynamic areas based on the complexity of the behavior. Subsequently, different degrees of enhancement are applied to dynamic areas with different levels of blur, so that the actions of people in the dynamic areas can be identified.

[0028] It should be further explained that the grayscale difference between the pixels in the current frame and the pixels at the same location in the previous frame is used to determine whether the pixels have changed. A larger grayscale difference indicates that the pixel location has changed due to the movement of people. Therefore, comparing the pixel differences between adjacent frames can accurately extract dynamic regions and provide basic data for subsequent behavior analysis and motion complexity assessment.

[0029] In this embodiment of the invention, each dynamic region of each frame of security inspection monitoring image is obtained: The preset difference threshold T=0.7. In other embodiments, the implementer can preset the value of the difference threshold according to the specific implementation situation. The absolute value of the grayscale difference between any pixel in the current frame security inspection monitoring image and the pixel at the same position in the previous frame security inspection monitoring image is recorded as the first difference. If the ratio of the first difference to 255 is greater than or equal to the difference threshold T, the pixel in the current frame security inspection monitoring image is recorded as a dynamic pixel. The region formed by all adjacent dynamic pixels in the current frame security inspection monitoring image is recorded as a dynamic region, thus obtaining each dynamic region of the current frame security inspection monitoring image. Similarly, each dynamic region of each frame of security inspection monitoring image is obtained.

[0030] It should be noted that complex actions such as people gathering, moving, and interacting occur in dynamic areas. The greater the complexity of the dynamic area's behavior, the more likely it is to become blurred. Therefore, optical flow is used to obtain the correspondence between dynamic areas in adjacent security monitoring images. If the area of ​​any dynamic area in the current frame of the security monitoring image changes significantly compared to the corresponding dynamic area in the previous frame, it indicates that complex actions such as people gathering, moving, and interacting have occurred in the dynamic area. In addition, the corner points of dynamic areas represent the body contour edges and joints (such as elbows, knees, and ankles) of people in the dynamic area. Therefore, the more corner points there are in any dynamic area in the current frame of the security monitoring image, the more complex actions such as people gathering, deforming, and interacting exist in the dynamic area. Based on the above characteristics, the complexity of the behavior of each dynamic area in the current frame of the security monitoring image is obtained.

[0031] In this embodiment of the invention, the matching relationship between the dynamic regions of the current frame security monitoring image and the previous frame security monitoring image is obtained by optical flow method, so as to obtain the dynamic region corresponding to each dynamic region of the current frame security monitoring image in the previous frame security monitoring image. The Harris corner detection algorithm is used to detect corners in each dynamic region of each frame of security monitoring image, thus obtaining the corners of each dynamic region of each frame of security monitoring image; Obtain the motion complexity of each dynamic region in the current frame of the security surveillance image: ; In the formula, This represents the behavioral motion complexity of the i-th dynamic region in the current frame of the security inspection monitoring image. This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image; This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image, corresponding to the dynamic region in the previous frame of the security inspection monitoring image. This represents the number of corner points in the i-th dynamic region of the current frame security monitoring image; norm() represents the normalization function. If the area of ​​a dynamic region changes significantly in adjacent frames, it means that complex actions such as people gathering, moving, or interacting may have occurred in the dynamic region. Therefore, the greater the complexity of the dynamic region's behavior, the more behaviors need to be identified, and the greater the difficulty of behavior recognition. Corner points represent the body contour edges and joints of a person, such as elbows, knees, and ankles. Therefore, if the number of corner points in the i-th dynamic region in the current frame of the security inspection image is more, it indicates that people in the dynamic region are engaging in complex actions such as gathering, deformation, or interaction.

[0032] It should be noted that the corner points in the dynamic region represent certain key feature points, such as the joints or contours of a person. By matching the corner points of the dynamic region in the current frame with the corresponding corner points in the previous frame, and calculating the Euclidean distance between each corner point in each dynamic region of the current security inspection image and its matching corner point, the movement of the corner points can be quantified, directly reflecting the distance the person moves in the image. Then, by obtaining the movement direction of each corner point in each dynamic region of the current security inspection image and its matching corner point, the movement direction of the corner points can be directly reflected. Therefore, the movement distance, movement direction, and position coordinates of each corner point in the dynamic region are used as the feature vector of each corner point, and different feature vectors represent different movement patterns. Therefore, by clustering the corner points of each dynamic region using the feature vectors of the corner points, multiple clusters are formed. Different clusters represent different motion behaviors within the dynamic region. Thus, the more clusters there are in the dynamic region and the more concentrated the distribution of all clusters in the dynamic region, the more diverse the types of motion behaviors in the dynamic region are, and the more concentrated the distribution of different motion behaviors is. At this time, the greater the chaos of motion behavior in the dynamic region.

[0033] In this embodiment of the invention, the SIFT (Scale-Invariant Feature Transform) algorithm is used to generate descriptors for corner points of each dynamic region in each frame of security monitoring image. Then, a feature matching algorithm is used to match corner points with similar descriptors in the corresponding dynamic region of the current frame of security monitoring image with corner points in the previous frame of security monitoring image, thereby obtaining the matching corner points of each corner point in each dynamic region of the current frame of security monitoring image. It should be noted that corner points that cannot be matched are not analyzed further.

[0034] The Euclidean distance between the position coordinates of the j-th corner point of the i-th dynamic region of the current frame security inspection image and the position coordinates of its matching corner point is denoted as the movement distance of the j-th corner point of the i-th dynamic region of the current frame security inspection image; the angle between the direction of the line connecting the position coordinates of the j-th corner point of the i-th dynamic region of the current frame security inspection image and its matching corner point and the horizontal direction is denoted as the movement direction of the j-th corner point of the i-th dynamic region of the current frame security inspection image. Using the moving distance, moving direction, and coordinate position of each corner point in each dynamic region of the current frame security inspection monitoring image as feature vectors, cluster all corner points in each dynamic region of the current frame security inspection monitoring image to obtain several clusters of each dynamic region of the current frame security inspection monitoring image; For any cluster in the i-th dynamic region of the current frame security inspection monitoring image, obtain the Euclidean distance between the cluster and all other clusters, and record the four nearest clusters as the neighboring clusters of the current cluster; Obtain the behavioral disorder of each dynamic region in the current frame of the security inspection surveillance image: ; In the formula, The behavioral disorder of the i-th dynamic region in the current frame of the security inspection and surveillance image; The Euclidean distance between the nth cluster of the i-th dynamic region of the current frame security monitoring image and the cluster center of its m-th neighboring cluster; The number of neighboring clusters of the nth cluster representing the i-th dynamic region of the current frame security inspection monitoring image; The number of clusters in the i-th dynamic region of the current frame security monitoring image; exp() represents the exponential function with the natural constant as the base; norm() represents the normalization function; The larger the value, the more movement behavior patterns of people in the i-th dynamic region of the current frame security inspection monitoring image, and the greater the behavioral chaos in the dynamic region; The larger the value, the closer each cluster in the i-th dynamic region of the current frame security inspection monitoring image is to its neighboring clusters. This indicates that all clusters in the dynamic region are concentrated, meaning that the distribution of different motion behaviors is more concentrated. In this case, the behavior disorder of the i-th dynamic region of the current frame security inspection monitoring image is greater.

[0035] It should be noted that the greater the complexity and disorder of the behavior in each dynamic area of ​​the current frame security monitoring image, the greater the complexity of the behavior in the dynamic area.

[0036] The product of the behavioral disorder of the i-th dynamic region in the current frame security inspection monitoring image and the behavioral motion complexity of the i-th dynamic region in the current frame security inspection monitoring image is taken as the behavioral complexity of the i-th dynamic region in the current frame security inspection monitoring image.

[0037] S3: Based on the behavioral complexity of each dynamic region in the current frame security inspection monitoring image, obtain the blur level of each dynamic region in the current frame security inspection monitoring image; based on the blur level of each dynamic region in the current frame security inspection monitoring image, obtain the enhancement coefficient of the current frame security inspection monitoring image.

[0038] It should be noted that the greater the behavioral complexity of any dynamic region in the current frame of the security inspection image, the more likely that dynamic region in the current frame of the security inspection image is to be blurred. Furthermore, when the behavioral complexity of the dynamic region continuously increases in consecutive frames of the security inspection image, it indicates that the actions of people in the dynamic region are becoming more frequent and the types of actions are more diverse. Therefore, the blurriness of the dynamic region in the current frame of the security inspection image is greater. Thus, based on the above characteristics, the blurriness of each dynamic region in the current frame of the security inspection image is obtained.

[0039] In this embodiment of the invention, the number of historical frame images is preset to G=5. In other embodiments, the implementer can preset the value of G according to the specific implementation situation. The matching relationship between the dynamic regions of the current frame security monitoring image and the previous G frame security monitoring images is obtained according to the optical flow method, so as to obtain the dynamic region corresponding to each dynamic region of the current frame security monitoring image in the previous G frame security monitoring images. The behavioral complexity of the i-th dynamic region of the current frame security monitoring image and the previous G frame security monitoring images is arranged according to the image acquisition order to obtain the behavioral complexity sequence of the i-th dynamic region of the current frame security monitoring image. Obtain the blur level of each dynamic region in the current frame of the security inspection surveillance image: ; In the formula, This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The behavior complexity of the i-th dynamic region in the current frame of the security inspection and surveillance image; The number of data points in the sequence representing the behavioral complexity of the i-th dynamic region of the current frame security monitoring image; The (d+1)th data value in the sequence of behavioral complexity of the i-th dynamic region of the current frame security inspection monitoring image; The d-th data value in the sequence of behavioral complexity of the i-th dynamic region of the current frame security monitoring image; norm() represents the normalization function; If the behavior of the i-th dynamic region in the current frame security monitoring image is more complex, it means that the dynamic region is more likely to become blurred, and the blur degree of the dynamic region is greater. This represents the increase in the complexity of the behavior of the i-th dynamic region in the continuous frame security inspection image. The larger the value, the more the complexity of the behavior of the i-th dynamic region in the continuous frame security inspection image increases. This means that the actions of people in the i-th dynamic region are becoming more frequent and the types of behavior are more diverse. Therefore, the blurriness of the i-th dynamic region in the current frame security inspection image is greater.

[0040] It should be noted that the greater the blurriness of each dynamic region in the current frame security inspection image, the greater the enhancement coefficient of the current frame security inspection image. Therefore, the enhancement coefficient of the current frame security inspection image is obtained based on the blurriness of all dynamic regions in the current frame security inspection image.

[0041] In this embodiment of the invention, the enhancement coefficient of the current frame security monitoring image is obtained: ; In the formula, This represents the enhancement factor of the current frame of the security surveillance image; This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The mean value representing the blur level of all dynamic regions in the current frame of the security inspection and surveillance image; This represents the number of dynamic regions in the current frame of the security inspection and monitoring image.

[0042] S4: Based on the enhancement coefficient of the current frame security inspection monitoring image, enhance the current frame security inspection monitoring image to obtain the current frame enhanced security inspection monitoring image.

[0043] It should be noted that the adaptive Laplacian sharpening algorithm is used to process the current frame of the security inspection image, enhancing the details and edges of the blurred image. The adaptive Laplacian sharpening algorithm assigns a sharpening intensity to each pixel in the image, with the value ranging from 0 to 1. The sharpening intensity determines the degree of blending between the original and sharpened images. When the sharpening intensity is 0, there is no enhancement effect, and the output image is the same as the original image. When the sharpening intensity is 1, the output image is entirely determined by the sharpened image, resulting in the strongest enhancement effect. Therefore, different values ​​between 0 and 1 represent different degrees of sharpening effect. Furthermore, since the closer the enhancement coefficient of the current frame of the security inspection image is to 1, the more blurred the image is, and the more enhancement is needed. Therefore, when the current frame of the security inspection image is highly blurred, the enhancement coefficient is close to 1, allowing for more sharpening processing.

[0044] In this embodiment of the invention, taking the current frame security inspection monitoring image as an example, the sharpening intensity of each pixel in the non-dynamic region of the current frame security inspection monitoring image is set to 0, and the enhancement coefficient of the current frame security inspection monitoring image is used as the sharpening intensity of each pixel in all dynamic regions of the current frame security inspection monitoring image to obtain the sharpening intensity of each pixel in the current frame security inspection monitoring image; based on the sharpening intensity of each pixel in the current frame security inspection monitoring image, the Laplacian sharpening algorithm is used to enhance the current frame security inspection monitoring image to obtain the enhanced image of the current frame security inspection monitoring image.

[0045] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for enhancing security inspection and surveillance images based on machine vision, characterized in that, include: Acquire each frame of security check surveillance image; obtain each dynamic region of each frame of security check surveillance image; Based on the changes in the area of ​​the dynamic region and the number of corner points, the behavioral motion complexity of each dynamic region in the current frame of the security inspection image is obtained; the feature vector of each corner point of each dynamic region in the current frame of the security inspection image is obtained; based on the feature vector, the dynamic regions are clustered for each corner point of the current frame of the security inspection image to obtain several clusters; based on the number and distribution of clusters, the behavioral disorder of each dynamic region in the current frame of the security inspection image is obtained; the behavioral motion complexity and behavioral disorder are multiplied to obtain the behavioral complexity of each dynamic region in the current frame of the security inspection image. Based on the complexity of the behavior, obtain the blur level of each dynamic region in the current frame of the security inspection monitoring image; Based on the degree of blur, the enhancement coefficient of the current frame security inspection image is obtained, and the current frame security inspection image is enhanced to obtain the current frame enhanced security inspection image.

2. The method for enhancing security inspection and monitoring images based on machine vision according to claim 1, characterized in that, The acquisition of each dynamic region of each frame of security monitoring image includes: A preset difference threshold T is used. The absolute value of the grayscale difference between any pixel in the current frame of the security monitoring image and the pixel at the same position in the previous frame of the security monitoring image is recorded as the first difference. If the ratio of the first difference to 255 is greater than or equal to the difference threshold T, the pixel in the current frame of the security monitoring image is recorded as a dynamic pixel. The region formed by all adjacent dynamic pixels in the current frame of the security monitoring image is recorded as a dynamic region, thus obtaining each dynamic region of the current frame of the security monitoring image.

3. The security inspection and monitoring image enhancement method based on machine vision according to claim 1, characterized in that, The acquisition of the behavioral motion complexity of each dynamic region in the current frame of the security inspection monitoring image includes: , This represents the behavioral motion complexity of the i-th dynamic region in the current frame of the security inspection monitoring image. This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image; This represents the area of ​​the i-th dynamic region in the current frame of the security inspection monitoring image, corresponding to the dynamic region in the previous frame of the security inspection monitoring image. This represents the number of corner points in the i-th dynamic region of the current frame security monitoring image; norm() represents the normalization function.

4. The security inspection and monitoring image enhancement method based on machine vision according to claim 1, characterized in that, The step of obtaining the feature vector of each corner point of each dynamic region in the current frame security inspection monitoring image includes: Obtain the matching corner points of each corner point in each dynamic region of the current frame security inspection monitoring image; The Euclidean distance between the position coordinates of the j-th corner point in the i-th dynamic region of the current frame security inspection image and the position coordinates of its matching corner point is denoted as the movement distance of the j-th corner point; the angle between the direction of the line connecting the position coordinates of the j-th corner point in the i-th dynamic region of the current frame security inspection image and its matching corner point and the horizontal direction is denoted as the movement direction of the j-th corner point; the movement distance, movement direction, and coordinate position of the j-th corner point in the i-th dynamic region of the current frame security inspection image are used as the feature vector of the j-th corner point.

5. The image enhancement method for security inspection monitoring based on machine vision according to claim 4, characterized in that, The step of obtaining the matching corner points of each corner point in each dynamic region of the current frame security inspection monitoring image includes: For each dynamic region of each frame of security inspection monitoring image, a SIFT descriptor is generated. Then, a feature matching algorithm is used to match the corner points of each dynamic region of the current frame of security inspection monitoring image with the corresponding dynamic region of the previous frame of security inspection monitoring image that have similar SIFT descriptors, so as to obtain the matching corner points of each corner point of each dynamic region of the current frame of security inspection monitoring image.

6. The image enhancement method for security inspection monitoring based on machine vision according to claim 1, characterized in that, The behavior of acquiring each dynamic region of the current frame security inspection monitoring image is chaotic, including: Obtain the neighboring clusters of each cluster in each dynamic region of the current frame security monitoring image; , The behavioral disorder of the i-th dynamic region in the current frame of the security inspection and surveillance image; The Euclidean distance between the nth cluster of the i-th dynamic region of the current frame security monitoring image and the cluster center of its m-th neighboring cluster; The number of neighboring clusters of the nth cluster representing the i-th dynamic region of the current frame security inspection monitoring image; The number of clusters in the i-th dynamic region of the current frame security monitoring image represents the number of clusters; exp() represents the exponential function with the natural constant as the base; norm() represents the normalization function.

7. The security inspection and monitoring image enhancement method based on machine vision according to claim 1, characterized in that, The process of obtaining the blur level of each dynamic region in the current frame security monitoring image includes: Obtain the sequence of behavioral complexity of the i-th dynamic region in the current frame of the security inspection monitoring image; , This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The behavior complexity of the i-th dynamic region in the current frame of the security inspection and surveillance image; The number of data points in the sequence representing the behavioral complexity of the i-th dynamic region of the current frame security monitoring image; The (d+1)th data value in the sequence of behavioral complexity of the i-th dynamic region of the current frame security inspection monitoring image; The d-th data value represents the sequence of behavioral complexity of the i-th dynamic region in the current frame of the security inspection monitoring image; norm() represents the normalization function.

8. The method for enhancing security inspection and monitoring images based on machine vision according to claim 1, characterized in that, The sequence of behavioral complexity for obtaining the i-th dynamic region of the current frame security monitoring image includes: The number of historical frame images is preset to G; the matching relationship between the dynamic regions of the current frame security monitoring image and the previous G frame security monitoring images is obtained according to the optical flow method, and the dynamic regions corresponding to each dynamic region of the current frame security monitoring image in the previous G frame security monitoring images are obtained; the behavioral complexity of the i-th dynamic region of the current frame security monitoring image and the previous G frame security monitoring images is arranged according to the image acquisition order, and the behavioral complexity sequence of the i-th dynamic region of the current frame security monitoring image is obtained.

9. The image enhancement method for security inspection monitoring based on machine vision according to claim 1, characterized in that, The step of obtaining the enhancement coefficient of the current frame security inspection monitoring image and enhancing the current frame security inspection monitoring image to obtain the current frame enhanced security inspection monitoring image includes: , This represents the enhancement factor of the current frame of the security monitoring image; This represents the blur level of the i-th dynamic region in the current frame of the security inspection monitoring image; The mean value representing the blur level of all dynamic regions in the current frame of the security inspection and surveillance image; This represents the number of dynamic regions in the current frame of the security inspection image. The enhancement coefficient of the current frame is used as the sharpening intensity of each pixel within all dynamic regions of the current frame, while the sharpening intensity of each pixel in the non-dynamic regions of the current frame is set to 0. Based on the sharpening intensity of each pixel in the current frame, the Laplacian sharpening algorithm is used to enhance the current frame, resulting in the enhanced security inspection image.

10. A security inspection and monitoring image enhancement system based on machine vision, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a machine vision-based security monitoring image enhancement method according to any one of claims 1-9.

Images