Traffic engineering construction site video monitoring method and system

Abstract

The invention relates to the technical field of image communication, in particular to a traffic engineering construction site video monitoring method and system, and the method comprises the steps: obtaining a monitoring frame image of each monitor at each moment in a traffic construction site, screening out the comparison monitor of each monitor at each moment, determining the risk coefficient of each macro block in the monitoring frame image, and determining the risk coefficient of each macro block in the monitoring frame image; a final comprehensive risk coefficient is determined, a plurality of key frame images of each monitor are screened out in combination with pixel point gray value differences between monitoring frame images of the same monitor at adjacent moments, each macro block in the key frame images is equally divided into a plurality of updated macro blocks, and the updated macro blocks are updated. And the compression coding module is used for compressing and coding all the monitored key frame images and transmitting all the monitored compressed data to the database. According to the invention, by introducing a risk assessment mechanism, the risk coefficients of different areas in the construction site monitoring picture are quantified, so that the compression effect of the video is monitored, and the bandwidth consumption is reduced.

Description

Technical Field

[0001] The present invention relates to the field of image communication technology, and in particular to a method and system for video monitoring of a traffic engineering construction site. Background Art

[0002] With the acceleration of urbanization, transportation construction projects are gradually increasing in the construction process of various roads and bridges. These construction projects are usually accompanied by high-risk working environments. Therefore, safety supervision of construction sites has become particularly important. In order to ensure safety during the construction process, a large number of monitoring equipment are usually deployed at transportation engineering construction sites to monitor the safety of the construction area in real time, especially in crowded areas and near construction equipment. However, due to the complex environment and large changes in the picture at the construction site, traditional video surveillance systems face huge bandwidth and storage pressure. How to effectively improve the quality of video surveillance and reduce bandwidth consumption in unimportant areas is a key technical problem in video surveillance at transportation construction sites.

[0003] Existing problems: In traditional video surveillance methods, the H.264 video encoding algorithm is widely used, but its macroblock division in the compression process mainly relies on the motion changes of the picture, while ignoring the risk differences in different areas of the construction site. Since there are a large number of areas with different risk levels on the construction site, the traditional encoding method cannot be optimized according to the importance of each area, resulting in excessive compression of unimportant areas, affecting the monitoring effect, and failing to effectively improve the clarity of important areas. Summary of the invention

[0004] The invention provides a traffic engineering construction site video monitoring method and system to solve the existing problems.

[0005] A traffic engineering construction site video monitoring method and system of the present invention adopts the following technical solutions: An embodiment of the present invention provides a method for video monitoring of a traffic engineering construction site, the method comprising the following steps: Obtain the monitoring frame image of each monitoring unit at each moment in the traffic construction site; According to the overlap between the monitoring frame images of different monitoring at the same time, the construction monitoring correlation between any two monitoring at each time is determined; according to the size of the construction monitoring correlation, the comparative monitoring of each monitoring at each time is screened out; The monitoring frame image is equally divided into several macroblocks; several target frames in the monitoring frame image are obtained, and the risk coefficient of each macroblock is determined according to the movement of the target frames in the monitoring frame image at different times of the same monitoring and the distribution of the target frames in adjacent macroblocks; According to the construction monitoring correlation between each monitoring and the comparison monitoring at each moment and the risk coefficient of each macroblock, the final comprehensive risk coefficient of each macroblock is determined; according to the final comprehensive risk coefficient, combined with the difference in pixel gray values ​​between monitoring frame images of the same monitoring at adjacent moments, a number of key frame images of each monitoring are screened out; Each macroblock in the key frame image is equally divided into a number of update macroblocks; according to the update macroblocks divided into the key frame images, all the key frame images of each monitoring are compressed and encoded to obtain the compressed data of each monitoring; and the compressed data of all monitoring are transmitted to the database.

[0006] Furthermore, the specific steps of determining the construction monitoring correlation between any two monitorings at each moment are as follows: In the At this moment, use the SIFT image registration algorithm to obtain the The monitoring frame image of the monitoring Overlapping area of ​​the monitoring frame images of the monitored , and then calculate the Monitoring and The sum of the areas of the monitored frames minus the The difference in area of The ratio of the area of ​​​​the first difference is recorded as Time next Monitoring and The intersection-over-union ratio of the overlapping areas of the monitoring frame images of the monitoring ; Get the and The intersection area ,in For the The monitoring frame image of the monitoring The overlapping area of ​​the monitoring frame images monitored; exist ,and Belong to the interval When inside, take all The maximum value in ,by is the correction factor, and the correction factor is The product of Time next Monitoring and Construction monitoring correlation between monitoring ;in, is the total number of monitoring, For the The area of For the area.

[0007] Furthermore, the specific steps of screening out the comparative monitoring of each monitoring at each moment include the following: When Time next Monitoring and Construction monitoring correlation between monitoring When it is greater than the preset judgment threshold, At this moment, The monitoring is recorded as A comparative monitoring of the monitoring.

[0008] Furthermore, the step of determining the risk coefficient of each macroblock includes the following specific steps: Obtain the target frame of the same object in each monitoring frame image at each moment; For Monitor and obtain the The monitoring frame image at the moment The object in the target box is The target frame in the monitoring frame image at the moment is recorded as the reference target frame; according to the center position coordinates of the reference target frame, the reference target frame points to the first The direction and Euclidean distance of the center position coordinates of the target frame constitute the first The moving speed vector corresponding to the target box ; In the Time next In the monitoring frame image of the monitoring The mean value of the modulus of the moving speed vectors corresponding to all target frames in the macroblock is recorded as The intensity of motion of objects within a macroblock , the The number of all target box center points in the macroblock is the same as the number of the The ratio of the area of ​​the first macroblock is used as the Object density in macroblock ; According to Time next The monitoring frame image of the monitoring The distribution of the target frame in the first macroblock and the adjacent macroblocks determines the The collision risk of macroblocks ; According to Time next The motion intensity, object density and collision risk of objects in all macroblocks in the monitoring frame image of the first monitoring are determined. The risk factor of each macroblock.

[0009] Furthermore, according to Time next The monitoring frame image of the monitoring The distribution of the target frame in the first macroblock and the adjacent macroblocks determines the The collision risk of macroblocks The specific steps are as follows: The said The center position coordinates of the target frame in the macroblock are recorded as the target coordinates. The coordinates of the center position of the target frame in all adjacent macroblocks of the macroblock are recorded as reference coordinates; Get the The moving speed vector of the target frame corresponding to the target coordinates , get the The target coordinates point to the The vector consisting of the direction and Euclidean distance of the reference coordinates , get the The reference coordinate points to the The vector consisting of the direction and Euclidean distance of the target coordinates , calculate the With the The cosine value of is taken as the first cosine value and the With the The cosine value of the first cosine value is taken as the second cosine value, the sum of the first cosine value and the second cosine value is calculated as the first sum value, and the first sum value is added to the The target coordinates to The ratio of the Euclidean distances of the reference coordinates is taken as the The target coordinates and The probability of collision between objects corresponding to the reference coordinates; Calculate the The sum of the probabilities of collision between the target coordinates and all objects corresponding to the reference coordinates is taken as the The collision factor of the target coordinates, the sum of the collision factors of all target coordinates is used as the first The collision risk of macroblocks .

[0010] Furthermore, according to Time next The motion intensity, object density and collision risk of objects in all macroblocks in the monitoring frame image of the first monitoring are determined. The risk coefficient of each macroblock includes the following specific steps: In the Time next In the monitoring frame image of the monitoring, the normalized value of the information entropy of the motion intensity of the objects in all macroblocks is used as the first weight, the normalized value of the information entropy of the density of the objects in all macroblocks is used as the second weight, and the normalized value of the information entropy of the collision risk of all macroblocks is used as the third weight. The first weight and the first weight are added together. The intensity of motion of objects within a macroblock The product of the second weight and the first Object density in macroblock The product of the third weight and the first The collision risk of macroblocks The product of the first product, the second product and the third product is taken as the third product, and the sum of the first product, the second product and the third product is taken as the first Risk factor of macroblock .

[0011] Furthermore, the determination of the final comprehensive risk coefficient of each macroblock includes the following specific steps: In the At this moment, from In all macroblocks of the monitoring frame images of all the monitoring comparison monitoring, SIFT image registration algorithm is used to obtain the Time next The monitoring frame image of the monitoring A number of matching macroblocks of a macroblock; Get the Time next Monitoring and The construction monitoring correlation between the contrast monitoring corresponding to the matching macroblocks The normalized value of , the With the said The product of the risk coefficients of the matching macroblocks is recorded as The comparison risk coefficients of matching macroblocks; among them, is a linear normalization function; The sum of the risk coefficients of all matching macroblocks and The sum of Time next The monitoring frame image of the monitoring The final comprehensive risk factor of the macroblock ;in, For the Time next The monitoring frame image of the monitoring The risk factor of each macroblock.

[0012] Furthermore, the specific steps of screening out a plurality of key frame images of each monitoring include the following: For monitoring, Moment and The sum of the absolute values ​​of the grayscale value differences of all pixels at the same position in the monitoring frame image at the time is taken as the The monitoring The amount of change in the monitoring frame image at the moment; Calculate the Time next The sum of the final comprehensive risk coefficients of all macroblocks in the monitoring frame image of the first monitoring is used as the second sum, and the second sum is added to the first The monitoring The product of the monitoring frame image changes at the time is taken as the The monitoring The monitoring frame image correction change at the moment; For A monitoring is performed in chronological order, with the monitoring frame image correction changes at all moments to form a monitoring frame image correction change time series sequence, and the monitoring frame images at the moments corresponding to all local maximum values ​​in the monitoring frame image correction change time series sequence are used as key frame images.

[0013] Furthermore, the step of equally dividing each macroblock in the key frame image into a plurality of update macroblocks includes the following specific steps: Preset Constants , get the The monitored The first key frame image The final comprehensive risk factor of the macroblock , the preset macroblock side length and The ratio of The monitored The first key frame image Updated side length of macroblocks ;in, is the floor function, is a linear normalization function; The first The monitored The first key frame image The macroblock is divided into several blocks of size Update macroblock.

[0014] The present invention also proposes a traffic engineering construction site video monitoring system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned traffic engineering construction site video monitoring method.

[0015] The beneficial effects of the technical solution of the present invention are: In an embodiment of the present invention, a monitoring frame image of each monitor at each moment in a traffic construction site is obtained, and the comparative monitoring of each monitor at each moment is screened out, and then the risk coefficient of each macroblock in the monitoring frame image of each monitor at each moment is determined, so as to determine the final comprehensive risk coefficient, and the difference in pixel gray values ​​between the monitoring frame images of the same monitor at adjacent moments is combined to screen out several key frame images of each monitor, thereby introducing a risk assessment mechanism to quantify the risk coefficients of different areas in the monitoring screen of the construction site, and adjust the quality of the video monitoring accordingly, and obtain high-quality key frame images for compression. Each macroblock in the key frame image is equally divided into several update macroblocks, which are used to compress and encode all the key frame images of each monitor, and obtain the compressed data of each monitor, thereby combining the risk assessment results in the video encoding process, and adaptively adjusting the size of the macroblock to adjust the high definition of the key area and the high compression ratio of the non-concern area. The compressed data of all monitors are transmitted to the database. At this point, the present invention improves the compression quality of surveillance videos, reduces bandwidth consumption, enhances the accuracy of security warnings, improves the flexibility of surveillance systems, and facilitates post-analysis and investigations, thereby having significant advantages in improving video surveillance compression quality, reducing bandwidth consumption, enhancing the accuracy of security warnings, improving the flexibility of surveillance systems, and facilitating post-analysis and investigations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, 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 work.

[0017] Figure 1 A flow chart of the steps of a method for video monitoring a traffic engineering construction site according to the present invention; Figure 2 This is a structural diagram of the traffic site construction monitoring system. DETAILED DESCRIPTION

[0018] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a traffic engineering construction site video monitoring method and system proposed by the present invention, its specific implementation method, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The specific scheme of a traffic engineering construction site video monitoring method and system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] See also Figure 1 , which shows a flow chart of the steps of a method for video monitoring a traffic engineering construction site provided by an embodiment of the present invention, the method comprising the following steps: Step S001: Acquire monitoring frame images of each monitoring unit at each moment in the traffic construction site.

[0022] It should be noted that at traffic construction sites, multiple surveillance cameras are usually installed to monitor the construction process from different angles to ensure construction safety. These cameras usually cover multiple directions and perspectives to capture possible safety risks, construction dynamics and personnel behavior in an all-round way.

[0023] For any traffic construction site, the monitoring frame images of each monitor at each moment are collected in real time.

[0024] It should be noted that: in this embodiment, all monitoring in the traffic construction site works simultaneously, collecting 30 frames of images per second, and a total of 20 monitoring cameras are installed. This is taken as an example for description, wherein the monitoring frame image is grayed to obtain the gray value of each pixel in the monitoring frame image. This is a well-known technology, and the specific method is not introduced here. In this embodiment, by building a complete video monitoring system, the monitoring video of the construction site is collected in real time, and data support is provided for subsequent security monitoring and video processing. In order to efficiently collect video information at the construction site, the system needs to deploy a series of equipment, including high-definition cameras, network transmission equipment, monitoring servers, and video processing terminals.

[0025] It should be further explained that the parameters of the various devices included in the system include: the resolution of the high-definition camera is 4K (3840×2160 pixels), the frame rate is 30fps (frames per second), and it has night vision function (built-in infrared light for night monitoring) and anti-interference ability (supports waterproof and dustproof level IP67, adapting to severe weather and environmental conditions). The bandwidth of the network transmission equipment supports data transmission bandwidth of more than 100Mbps (megabits per second), and the protocol supports Ethernet, Wi-Fi and 5G (the appropriate transmission method can be selected according to the on-site conditions). The CPU of the video surveillance server is Intel Xeon Gold 6240R 2.4 GHz, the memory is 64GB RAM, the hard disk is 2TB SSD (for efficient storage of video data), and the video decoding card is NVIDIA Tesla T4 to support 4K video decoding and real-time video processing. The display of the monitoring center processing terminal is a 4K display with a display size of 32 inches and an operating system of Windows 10 (supporting video processing software and monitoring platform). In this embodiment, the structural diagram of the traffic site construction monitoring system is as follows: Figure 2 shown. Figure 2 After the monitoring equipment obtains the construction site image, the monitoring terminal compresses the video data through compression encoding and sends it to the communication base station. After receiving the transmitted data through the monitoring management platform, it decodes it and sends the processed information to the monitoring center through the Internet and private network to realize real-time video monitoring of the traffic construction site.

[0026] Step S002: Determine the construction monitoring correlation between any two monitors at each moment according to the overlap between the monitoring frame images of different monitors at the same moment; and select the comparative monitoring of each monitor at each moment according to the size of the construction monitoring correlation.

[0027] It should be noted that: at traffic construction sites, multiple surveillance cameras are usually installed to monitor the construction process from different angles to ensure construction safety. Therefore, for the same area photographed from different angles, the SIFT image registration algorithm is used to obtain the image overlap area between multiple surveillance videos. Among them, the SIFT (Scale-Invariant Feature Transform) image registration algorithm is a well-known technology, and the specific method is not introduced here.

[0028] Preferably, in one embodiment of the present invention, the method for obtaining the comparative monitoring of each monitoring at each moment includes: First For example, in the At this moment, use the SIFT image registration algorithm to obtain the The monitoring frame image of the monitoring Overlapping area of ​​the monitoring frame images of the monitored , and then calculate the Monitoring and The sum of the areas of the monitored frames minus The difference in area of ​​​​ The ratio of the area of ​​​​the first difference is recorded as Time next Monitoring and The intersection-over-union ratio of the overlapping areas of the monitoring frame images of the monitoring .in, .

[0029] According to the above method, get Time next The intersection-and-union ratio of the overlapping area of ​​the monitoring frame image of a monitoring unit with that of each other monitoring unit.

[0030] What needs to be explained is that it is necessary to continue to analyze the remaining monitoring and The overlapping area between the monitoring areas of the first monitoring is selected Multiple monitorings with strong correlation between them are analyzed The possible risk situations in different areas of the monitoring screen. In the content of each monitoring screen, there may be some areas without contrast images or with small contrast images. Therefore, the relationship between the overlapping areas of the surveillance frame images of different monitors is considered to obtain the overlap between each monitor and the first The construction monitoring correlation between the monitoring screens.

[0031] Get and The intersection area ,in For the The monitoring frame image of the monitoring The overlapping area of ​​the surveillance frame images of the two monitored devices.

[0032] exist ,and Belong to the interval When inside, take all The maximum value in ,by is the correction factor, and the correction factor is The product of Time next Monitoring and Construction monitoring correlation between monitoring .in, is an integer, is the total number of monitoring, for The area of for area.

[0033] What needs to be explained is: The larger the Time next The first The larger the overlapping area of ​​the two monitoring When smaller, Time next The first The overlapping area of ​​the first monitoring The intersection of the first and other monitoring overlapping areas is small, that is, The first The higher the uniqueness of the overlapping area of ​​two monitorings, the better the correlation between the two monitorings. for The correction factor is obtained .

[0034] The preset judgment threshold is 0.5, and this is used as an example for description.

[0035] when When it is greater than the preset judgment threshold, At this moment, monitoring, denoted as A comparative monitoring of the monitoring.

[0036] According to the above method, get At this moment, All comparison monitoring of a monitoring.

[0037] What needs to be explained is: according to the correlation of construction monitoring, When compressing and transmitting a surveillance picture, multiple surveillance angles with high correlation are selected for comparative analysis to improve the reliability of subsequent compressed video transmission for high-risk areas.

[0038] Step S003: Divide the monitoring frame image into several macroblocks; obtain several target frames in the monitoring frame image, and determine the risk coefficient of each macroblock according to the movement of the target frames in the monitoring frame image at different times of the same monitoring and the distribution of the target frames in adjacent macroblocks.

[0039] It should be noted that: when transmitting video images, it is necessary to compress the video at the monitoring terminal. Before compression, the traditional H.264 compression coding needs to divide the image into macroblocks, and generally uses 16×16 pixel macroblocks as the basic coding unit. Among them, H.264 compression coding is a well-known technology, and the specific method is not introduced here. When real-time transmission and detection of monitoring video is performed, the transmitted video segment is the compressed video data between two key frames. Therefore, after the previous video segment data is compressed and transmitted to the monitoring management platform, it is considered that after the management platform decodes the received video, the objects in the video image are identified by the traditional YOLO algorithm, specifically including vehicles, pedestrians, etc. The recognition result of the YOLO algorithm includes the target frame position of the corresponding object in each frame image and the confidence size of the corresponding target frame. For the video data within a period of time after transmission, the target frame will move in the image as the object (personnel, vehicle) moves. Among them, the YOLO algorithm is a target detection model, which is a well-known technology, and the specific method is not introduced here.

[0040] Preferably, in one embodiment of the present invention, a method for obtaining a risk coefficient of each macroblock in each monitored monitoring frame image at each moment includes: Use the YOLO algorithm to obtain several target boxes in the monitoring frame image of each monitor at each moment.

[0041] Use the target tracking algorithm to obtain the target frame of the same object in each monitoring frame image at each moment.

[0042] Among them, the target tracking algorithm selects the optical flow method, which is a well-known technology and the specific method will not be introduced here.

[0043] First For example, for the first Monitor and obtain the The monitoring frame image at the moment The object in the target box is The target frame in the monitoring frame image at the moment is recorded as the reference target frame. According to the coordinates of the center position of the reference target frame, point to the The direction and Euclidean distance of the center coordinates of the target frame constitute the The moving speed vector corresponding to the target box .

[0044] It should be noted that if there is no reference target frame, the subsequent analysis does not include the first The monitoring The monitoring frame image at the moment The monitoring frame image at the first moment is not analyzed subsequently. The lower left corner vertex of the monitoring frame image is taken as the origin, the horizontal right is the horizontal axis square, and the vertical upward is the vertical axis square to construct a coordinate system, so as to obtain the coordinates of the center position of the target frame. Therefore, after the management platform receives the video segments sent back by multiple monitoring, it obtains the movement speed vectors of different objects (target frames) in each monitoring according to the object recognition results in multiple video segments.

[0045] The default macroblock side length is , taking this as an example for description.

[0046] Divide each monitoring frame image at each moment into several equal parts of size macroblocks.

[0047] In the Time next Among the monitored frames, count the The target box exists within the macroblock.

[0048] It should be noted that part of the target frame is within the macroblock, and is also the target frame within the macroblock. The moving speeds corresponding to the multiple target frames within the macroblock are different, and in order to analyze the risk factors of different macroblocks, it is necessary to analyze the movement of objects within different macroblocks.

[0049] In the Time next In the monitoring frame image of the monitoring The mean value of the modulus of the moving speed vectors corresponding to all target frames in the macroblock is recorded as The intensity of motion of objects within a macroblock , will The number of all target box center points in the macroblock is The ratio of the area of ​​the first macroblock is used as the Object density in macroblock .

[0050] It should be noted that construction safety is particularly important when monitoring a construction site. Therefore, based on the above characteristics, it is considered to obtain the collision risk of macroblocks by analyzing the center point positions of multiple objects and the correlation between their movement speed vectors.

[0051] In the Time next In the monitoring frame image of the monitoring The center coordinates of the target frame in the macroblock are recorded as the target coordinates. The coordinates of the center position of the target box in all adjacent macroblocks of a macroblock are recorded as reference coordinates.

[0052] Get the The moving speed vector of the target frame corresponding to the target coordinates , get the The target coordinates point to the The vector consisting of the direction and Euclidean distance of the reference coordinates , get the The reference coordinate points to the The vector consisting of the direction and Euclidean distance of the target coordinates ,calculate and The cosine value of is used as the first cosine value to calculate and The cosine value of the first cosine value is taken as the second cosine value, and the sum of the first cosine value and the second cosine value is calculated as the first sum value. The target coordinates to The ratio of the Euclidean distances of the reference coordinates is taken as the The target coordinates and The probability of collision between objects corresponding to the reference coordinates.

[0053] It should be noted that the cosine value of two vectors is a method to measure the directional similarity of two vectors in multidimensional space. This is a well-known technology and the specific method is not introduced here. The larger the cosine value, the more the two vectors are in the same direction, and the smaller the clustering of the two coordinates, the more likely a collision will occur.

[0054] Calculate the The sum of the probabilities of collision between the target coordinates and all objects corresponding to the reference coordinates is taken as the The collision factor of the target coordinates, the sum of the collision factors of all target coordinates is taken as the Time next The monitoring frame image of the monitoring The collision risk of macroblocks .

[0055] In the Time next In the monitoring frame image of the monitoring, the normalized value of the information entropy of the motion intensity of the objects in all macroblocks is used as the first weight, the normalized value of the information entropy of the density of the objects in all macroblocks is used as the second weight, and the normalized value of the information entropy of the collision risk of all macroblocks is used as the third weight. The first weight and the third weight are added together. The intensity of motion of objects within a macroblock The product of is taken as the first product, and the second weight is multiplied by the Object density in macroblock The product of the third weight and the The collision risk of macroblocks The product of the first product, the second product and the third product is taken as the third product. Risk factor of macroblock .

[0056] It should be noted that information entropy is a measure of information uncertainty. It describes the uncertainty of the value of a random variable. This is a well-known technology and the specific method is not introduced here. The larger the information entropy, the higher the uncertainty of the information source, that is, the probability of occurrence of each possible event in the information source is more uniform. In this case, the amount of information carried by each event is relatively large, because their occurrence is relatively unexpected, so it is given a larger weight. In this embodiment, The linear normalization function normalizes the information entropy.

[0057] according to The risk coefficient of each macroblock in each monitored frame image at each moment is obtained by the acquisition method.

[0058] Step S004: Determine the final comprehensive risk coefficient of each macroblock based on the construction monitoring correlation between each monitoring and the comparison monitoring at each moment and the risk coefficient of each macroblock; based on the final comprehensive risk coefficient, combined with the difference in pixel grayscale values ​​between monitoring frame images of the same monitoring at adjacent moments, screen out several key frame images of each monitoring.

[0059] It should be noted that in the monitoring images at each corresponding monitoring angle, the shooting angle may cause errors in the target recognition results due to occlusion and other reasons, resulting in a smaller number of target frames detected in some macroblocks, which in turn leads to a smaller risk factor.

[0060] Preferably, in one embodiment of the present invention, the method for acquiring a plurality of key frame images for each monitoring comprises: In the At this moment, from In all macroblocks of the monitoring frame images of all the monitoring comparison monitoring, SIFT image registration algorithm is used to obtain the Time next The monitoring frame image of the monitoring A number of matching macroblocks of a macroblock.

[0061] Get the Time next Monitoring and The construction monitoring correlation between the contrast monitoring corresponding to the matching macroblocks The normalized value of ,Will With The product of the risk coefficients of the matching macroblocks is recorded as The comparison risk coefficients of matching macroblocks. is a linear normalization function.

[0062] The sum of the risk coefficients of all matching macroblocks and The sum of Time next The monitoring frame image of the monitoring The final comprehensive risk factor of the macroblock .in, For the Time next The monitoring frame image of the monitoring The risk factor of each macroblock.

[0063] It should be noted that: according to the risk coefficient size of each macroblock, when transmitting the monitoring image, it is necessary to consider adjusting the selection of the next key frame. In traditional H.264 encoding, the selection of key frames is usually based on fixed intervals or specific scene changes. However, this method is not always able to effectively reflect the actual risk distribution of the construction site, resulting in a certain decrease in the clarity of the video images in high-risk areas. The traditional H.264-based scene change detection method calculates the sum of the grayscale changes between two adjacent frames, and judges the key frames based on a fixed threshold.

[0064] For monitoring, Moment and The sum of the absolute values ​​of the grayscale value differences of all pixels at the same position in the monitoring frame image at the time is taken as the The monitoring The amount of change in the monitoring frame image at the moment.

[0065] Calculate the Time next The sum of the final comprehensive risk coefficients of all macroblocks in the monitoring frame image of the first monitoring is taken as the second sum value, and the second sum value is added to the first The monitoring The product of the monitoring frame image changes at the time is taken as the The monitoring The monitoring frame image correction change at the moment.

[0066] For A monitoring is performed in chronological order, with the monitoring frame image correction changes at all moments to form a monitoring frame image correction change time series sequence, and the monitoring frame images at the moments corresponding to all local maximum values ​​in the monitoring frame image correction change time series sequence are used as key frame images.

[0067] It should be noted that: in this embodiment, a peak detection algorithm is used to obtain the local maximum value in the time series, which is a well-known technology and the specific method is not introduced here. In this embodiment, the key frame image analysis is not performed on the monitoring frame image at the first moment.

[0068] According to the above method, several key frame images of each monitoring are obtained.

[0069] Step S005: Divide each macroblock in the key frame image into a number of update macroblocks; compress and encode all key frame images of each monitoring according to the update macroblocks divided by the key frame images to obtain compressed data of each monitoring; and transmit the compressed data of all monitoring to the database.

[0070] It should be noted that after determining the timing of key frame selection, the system will further adjust the size of the macroblocks, and refine the macroblocks in the high-risk area according to the risk coefficient of the macroblocks, that is, use smaller macroblocks for encoding to maintain higher clarity. For low-risk areas, the size of the macroblocks can be appropriately increased to improve compression efficiency.

[0071] Preferably, in one embodiment of the present invention, the method for acquiring compressed data of each monitoring includes: Preset Constants , taking this as an example for description.

[0072] Get the The monitored The first key frame image The final comprehensive risk factor of the macroblock , the preset macroblock side length and The ratio of The monitored The first key frame image Updated side length of macroblocks .in, is the floor function, is a linear normalization function.

[0073] It should be noted that: in this embodiment The values ​​of are 16, 8 and 4 respectively. According to different values, the initial 16×16 macroblock size remains unchanged or is divided into multiple 8×8 or 4×4 sub-blocks for subsequent compression.

[0074] The first The monitored The first key frame image The macroblock is divided into several blocks of size Update macroblock.

[0075] According to the above method, a number of updated macroblocks equally divided from each macroblock in each monitored key frame image are obtained.

[0076] According to the updated macroblocks divided by each key frame image of each monitoring, H.264 compression coding is used to compress and encode all the key frame images of each monitoring, obtain the compressed data of each monitoring, and transmit the compressed data of all monitoring to the database.

[0077] It should be noted that: according to the adjusted update macroblock size, the key frame image is compressed and transmitted to the database of the monitoring management platform through the network. In this way, even when the network bandwidth is limited, the video quality of the key area can be guaranteed while reducing the overall bandwidth consumption. After completing the video acquisition and update macroblock division, the system will enter the real-time monitoring stage. The operator of the monitoring center can view the situation of the construction site in real time through the monitoring terminal, and the operator of the monitoring center can manually adjust the display ratio and details of the monitoring screen according to the actual situation, so as to better observe the safety situation of the construction site. The monitoring system can also record the historical data of all monitoring screens for subsequent analysis and investigation.

[0078] It should be noted that: in this embodiment, when the ratio is obtained, when the denominator is 0, the denominator is set to 1 to ensure that the ratio is established.

[0079] The present invention also provides a traffic engineering construction site video monitoring system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned traffic engineering construction site video monitoring method.

[0080] So far, the present invention is completed.

[0081] In summary, in an embodiment of the present invention, a monitoring frame image of each monitor at each moment in a traffic construction site is obtained, and the comparative monitoring of each monitor at each moment is screened out, and then the risk coefficient of each macroblock in the monitoring frame image of each monitor at each moment is determined, so as to determine the final comprehensive risk coefficient, and the difference in pixel gray values ​​between the monitoring frame images of the same monitor at adjacent moments is combined to screen out several key frame images of each monitor, and each macroblock in the key frame image is equally divided into several update macroblocks, so as to compress and encode all key frame images of each monitor, obtain the compressed data of each monitor, and transmit the compressed data of all monitors to the database. The present invention realizes the quantification of the risk coefficients of different areas in the monitoring screen of the construction site by introducing a risk assessment mechanism, so as to monitor the compression effect of the video and reduce bandwidth consumption.

[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for video monitoring of a traffic engineering construction site, characterized in that: The method comprises the following steps: Obtain the monitoring frame image of each monitoring unit at each moment in the traffic construction site; According to the overlap between the monitoring frame images of different monitoring at the same time, the construction monitoring correlation between any two monitoring at each time is determined; according to the size of the construction monitoring correlation, the comparative monitoring of each monitoring at each time is screened out; The monitoring frame image is equally divided into several macroblocks; several target frames in the monitoring frame image are obtained, and the risk coefficient of each macroblock is determined according to the movement of the target frames in the monitoring frame image at different times of the same monitoring and the distribution of the target frames in adjacent macroblocks; According to the construction monitoring correlation between each monitoring and the comparison monitoring at each moment and the risk coefficient of each macroblock, the final comprehensive risk coefficient of each macroblock is determined; according to the final comprehensive risk coefficient, combined with the difference in pixel gray values ​​between monitoring frame images of the same monitoring at adjacent moments, a number of key frame images of each monitoring are screened out; Each macroblock in the key frame image is equally divided into a number of update macroblocks; according to the update macroblocks divided into the key frame images, all the key frame images of each monitoring are compressed and encoded to obtain the compressed data of each monitoring; and the compressed data of all monitoring are transmitted to the database.

2. According to claim 1, a method for video monitoring of a traffic engineering construction site is characterized in that: The specific steps of determining the construction monitoring correlation between any two monitorings at each moment are as follows: In the At this moment, use the SIFT image registration algorithm to obtain the The monitoring frame image of the monitoring Overlapping area of ​​the monitoring frame images of the monitored , and then calculate the Monitoring and The sum of the areas of the monitored frames minus the The difference in area of The ratio of the area of ​​​​the first difference is recorded as Time next Monitoring and The intersection-over-union ratio of the overlapping area of ​​the monitoring frame images of the monitoring ; Get the and The intersection area ,in For the The monitoring frame image of the monitoring The overlapping area of ​​the monitoring frame images monitored; exist ,and Belong to the interval When inside, take all The maximum value in ,by is the correction factor, and the correction factor is The product of Time next Monitoring and Construction monitoring correlation between monitoring ;in, is the total number of monitoring, For the The area of For the area.

3. A method for video monitoring of a traffic engineering construction site according to claim 1, characterized in that: The specific steps of screening out the comparative monitoring of each monitoring at each moment are as follows: When Time next Monitoring and Construction monitoring correlation between monitoring When it is greater than the preset judgment threshold, At this moment, The monitoring is recorded as A comparative monitoring of the monitoring.

4. A method for video monitoring of a traffic engineering construction site according to claim 1, characterized in that: The specific steps of determining the risk coefficient of each macroblock are as follows: Obtain the target frame of the same object in each monitoring frame image at each moment; For Monitor and obtain the The monitoring frame image at the moment The object in the target box is The target frame in the monitoring frame image at the moment is recorded as the reference target frame; according to the center position coordinates of the reference target frame, the reference target frame points to the first The direction and Euclidean distance of the center position coordinates of the target frame constitute the first The moving speed vector corresponding to the target box ; In the Time next In the monitoring frame image of the monitoring The mean value of the modulus of the moving speed vectors corresponding to all target frames in the macroblock is recorded as The intensity of motion of objects within a macroblock , the The number of all target box center points in the macroblock is the same as the number of the The ratio of the area of ​​the first macroblock is used as the Object density in macroblock ; According to Time next The monitoring frame image of the monitoring The distribution of the target frame in the first macroblock and the adjacent macroblocks determines the The collision risk of macroblocks ; According to Time next The motion intensity, object density and collision risk of objects in all macroblocks in the monitoring frame image of the first monitoring are determined. The risk factor of each macroblock.

5. A method for video monitoring of a traffic engineering construction site according to claim 4, characterized in that: According to the Time next The monitoring frame image of the monitoring The distribution of the target frame in the first macroblock and the adjacent macroblocks determines the The collision risk of macroblocks The specific steps are as follows: The said The center position coordinates of the target frame in the macroblock are recorded as the target coordinates. The coordinates of the center position of the target frame in all adjacent macroblocks of the macroblock are recorded as reference coordinates; Get the The moving speed vector of the target frame corresponding to the target coordinates , get the The target coordinates point to the The vector consisting of the direction and Euclidean distance of the reference coordinates , get the The reference coordinate points to the The vector consisting of the direction and Euclidean distance of the target coordinates , calculate the With the The cosine value of is taken as the first cosine value and the With the The cosine value of the first cosine value is taken as the second cosine value, the sum of the first cosine value and the second cosine value is calculated as the first sum value, and the first sum value is added to the The target coordinates to The ratio of the Euclidean distances of the reference coordinates is taken as the The target coordinates and The probability of collision between objects corresponding to the reference coordinates; Calculate the The sum of the probabilities of collision between the target coordinates and all objects corresponding to the reference coordinates is taken as the The collision factor of the target coordinates, the sum of the collision factors of all target coordinates is used as the first The collision risk of macroblocks .

6. A method for video monitoring of a traffic engineering construction site according to claim 4, characterized in that: According to the Time next The motion intensity, object density and collision risk of objects in all macroblocks in the monitoring frame image of the first monitoring are determined. The risk coefficient of each macroblock includes the following specific steps: In the Time next In the monitoring frame image of the monitoring, the normalized value of the information entropy of the motion intensity of the objects in all macroblocks is used as the first weight, the normalized value of the information entropy of the density of the objects in all macroblocks is used as the second weight, and the normalized value of the information entropy of the collision risk of all macroblocks is used as the third weight. The first weight is added to the first weight. The intensity of motion of objects within a macroblock The product of the second weight and the first Object density in macroblock The product of the third weight and the first The collision risk of macroblocks The product of the first product, the second product and the third product is taken as the third product, and the sum of the first product, the second product and the third product is taken as the first Risk factor of macroblock .

7. A method for video monitoring of a traffic engineering construction site according to claim 1, characterized in that: The specific steps of determining the final comprehensive risk coefficient of each macroblock are as follows: In the At this moment, from In all macroblocks of the monitoring frame images of all the monitoring comparison monitoring, SIFT image registration algorithm is used to obtain the Time next The monitoring frame image of the monitoring A number of matching macroblocks of a macroblock; Get the Time next Monitoring and The construction monitoring correlation between the contrast monitoring corresponding to the matching macroblocks The normalized value of , the With the said The product of the risk coefficients of the matching macroblocks is recorded as The comparison risk coefficients of matching macroblocks; among them, is a linear normalization function; The sum of the risk coefficients of all matching macroblocks and The sum of Time next The monitoring frame image of the monitoring The final comprehensive risk factor of the macroblock ;in, For the Time next The monitoring frame image of the monitoring The risk factor of each macroblock.

8. A method for video monitoring of a traffic engineering construction site according to claim 1, characterized in that: The specific steps of screening out a plurality of key frame images for each monitoring are as follows: For monitoring, Moment and The sum of the absolute values ​​of the grayscale value differences of all pixels at the same position in the monitoring frame image at the time is taken as the The monitoring The amount of change in the monitoring frame image at the moment; Calculate the Time next The sum of the final comprehensive risk coefficients of all macroblocks in the monitoring frame image of the first monitoring is used as the second sum, and the second sum is added to the first The monitoring The product of the monitoring frame image changes at the time is taken as the The monitoring The monitoring frame image correction change at the moment; For A monitoring is performed in chronological order, with the monitoring frame image correction changes at all moments to form a monitoring frame image correction change time series sequence, and the monitoring frame images at the moments corresponding to all local maximum values ​​in the monitoring frame image correction change time series sequence are used as key frame images.

9. A method for video monitoring of a traffic engineering construction site according to claim 1, characterized in that: The specific steps of equally dividing each macroblock in the key frame image into a plurality of update macroblocks are as follows: Preset Constants , get the The monitored The first key frame image The final comprehensive risk factor of the macroblock , the preset macroblock side length and The ratio of The monitored The first key frame image Updated side length of macroblocks ;in, is the floor function, is a linear normalization function; The first The monitored The first key frame image The macroblock is divided into several blocks of size Update macroblock.

10. A traffic engineering construction site video monitoring system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of a method for video monitoring a traffic engineering construction site as described in any one of claims 1 to 9 are implemented.

Images