CN-122002001-A - Dual-prevention-based illegal behavior monitoring equipment and method

CN122002001ACN 122002001 ACN122002001 ACN 122002001ACN-122002001-A

Abstract

The invention relates to the field of illegal behavior monitoring, in particular to illegal behavior monitoring equipment and method based on double prevention, which are used for solving the technical problems that the illegal behavior monitoring method in the prior art is lack of predictive analysis and risk prevention and control mechanisms, so that the early warning capability and prevention and control effect of a monitoring system are limited, and a passive response mode is mostly adopted, namely, illegal behaviors can not be found and recorded in advance only after the occurrence of the illegal behaviors, and cannot be early warned and prevented.

Inventors

Chai rui
LIU JUN
HE ZHANYOU
LIU XIANGHAO
CHEN ZHIYONG
ZHONG DING
SUN CHENG
TONG JUNFENG

Assignees

中国石油天然气股份有限公司

Dates

Publication Date: 20260508
Application Date: 20241108

Claims (10)

1. The dual-prevention-based illegal behavior monitoring equipment comprises a high-definition camera (1) and is characterized by further comprising a mounting frame (2) and a central processing unit (3), wherein the mounting frame (2) is arranged on one side of the high-definition camera (1), the central processing unit (3) is arranged on the inner side of the high-definition camera (1), the mounting frame (2) is used for mounting the high-definition camera (1) at a key point position to be monitored, the high-definition camera (1) is used for capturing high-definition images and video information, and the central processing unit (3) is used for analyzing and identifying illegal behaviors in videos and images in real time by adopting image processing and machine learning algorithms.
2. The double prevention-based violation monitoring method according to claim 1, wherein: s1, presetting behavior characteristics and risk grades, and carrying out risk grade division according to the categories of different behavior characteristics; s2, acquiring high-definition video image data in real time through a high-definition camera (1); S3, carrying out graying, noise reduction and enhancement pretreatment on video image data captured by the high-definition camera (1); s4, extracting characteristic information such as edges, textures, shapes and the like from the preprocessed image; s5, classifying and identifying the extracted static behavior characteristics; s6, classifying and identifying the extracted dynamic behavior characteristics; S7, carrying out risk assessment on the illegal behavior according to the nature and the result of the illegal behavior to determine the hazard level of the illegal behavior, outputting the identification result to related personnel in the form of images, videos or alarm information, and taking corresponding treatment measures aiming at the identified hidden danger; S8, predicting whether the behavior features of the static behavior features and the dynamic behavior features which are not recognized as risk behaviors are likely to generate risks after a period of time by a machine learning algorithm; And S9, outputting a prediction result to related personnel, and making risk prevention and control measures in advance for predicting possible risk behaviors.
3. The method for monitoring the illegal behaviors based on double prevention according to claim 2 is characterized by comprising the steps of presetting behavior characteristics and risk levels, classifying the risk levels according to the categories of different behavior characteristics, determining the category of the behavior characteristics to be monitored, classifying the behavior characteristics into different risk levels according to the severity of consequences possibly caused by each behavior characteristic, and establishing a mapping relation between the behavior characteristics and the risk levels, wherein the method comprises the following specific steps: s101, setting a monitoring target and a monitoring range, namely determining which types of behavior features need to be monitored; S102, taking industry standards and specifications as the basis for defining behavior characteristics and classifying risk levels; s103, creating an exhaustive list to list all behavior feature categories to be monitored; s104, according to analysis of related laws and regulations, industry standards and historical accident cases, formulating standards of risk classification, wherein the standards relate to the severity degree, occurrence frequency and influence range of consequences possibly caused by behavior characteristics; s105, dividing behavior characteristics into different risk levels according to established standards; s106, providing an explicit description for each risk level, including consequences and countermeasures which can be caused by the explicit description; S107, establishing a mapping relation table of behavior characteristics and risk levels; s108, inputting the defined behavior characteristics and the divided risk levels into a mapping relation table, wherein each behavior characteristic is associated with a specific risk level; And S109, verifying the mapping relation table to ensure that all behavior characteristics are correctly classified and corresponding risk levels are allocated.
4. The dual-prevention-based illegal behavior monitoring method according to claim 3 is characterized by ensuring that the installation position of a camera can fully cover a monitoring area and is adjusted to an optimal shooting angle, acquiring high-definition video image data in real time through a high-definition camera (1) and transmitting the high-definition video image data to a monitoring center, wherein the method comprises the following specific steps: s201, determining the required number of cameras and mounting positions according to the monitoring purpose; S202, installing a camera and adjusting the angle of the camera so that details of people, vehicles, objects and the like in a target area can be clearly captured; S203, adjusting parameters such as focal length, brightness, contrast and the like to enable the picture to achieve the best effect; S204, directly connecting the camera with a monitoring center in a wired mode such as a network cable; S205, the original video data collected by the camera needs to be encoded; s206, selecting TCP/IP transmission protocol according to network environment and monitoring requirement; and S207, the camera transmits the encoded and compressed video data to the monitoring center in real time.
5. The double-prevention-based illegal behavior monitoring method according to claim 4, wherein the method is characterized by carrying out graying, noise reduction and enhancement pretreatment on video image data captured by a high-definition camera (1), converting a color video image into a gray image through graying, reducing calculated amount, removing image noise through noise reduction, improving image quality through an enhancement technology, and facilitating subsequent feature extraction, and comprises the following specific steps: S301, reading a frame of color image from a video stream, wherein the image consists of three color channels of red (R), green (G) and blue (B); S302, giving R, G, B different weights (R: 0.299, G:0.587, B: 0.114) according to the sensitivity of human eyes to different colors, and calculating a weighted average value as a gray value; S303, calculating the gray value of each pixel according to a weighted average method; s304, giving the calculated gray value to each pixel to generate a gray image; s305, carrying out weighted average on the image through a Gaussian function, smoothing the image and removing noise; S306, checking the noise-reduced image to ensure that the noise is effectively suppressed, and meanwhile, the details of the image are not excessively blurred; s307, by adjusting the histogram distribution of the image, the pixel value distribution of the image is more uniform, and the contrast of the image is improved; S308, by stretching the contrast range of the image, the dark part of the image is darker, the bright part is brighter, and the visual effect of the image is improved.
6. The method for monitoring the illegal behaviors based on double prevention according to claim 5, wherein the method is characterized by extracting characteristic information such as edges, textures, shapes and the like from the preprocessed image, extracting the edge information of the image through an edge detection algorithm, extracting the texture characteristics of the image through a texture analysis method, and identifying the shape characteristics in the image through a contour detection algorithm, and comprises the following specific steps: S401, selecting a Canny edge detection algorithm according to application requirements; s402, calculating the amplitude and direction of the image gradient; s403, refining edges by applying non-maximum suppression; S404, detecting and connecting edges by using a double-threshold algorithm; S405, selecting a gray level co-occurrence matrix GLCM texture analysis method according to application requirements; S406, defining a distance d and a direction theta; s407, counting the frequency of gray level joint distribution between adjacent pixel points with the distance d and the direction theta for each pixel point in the image; S408, constructing a gray level co-occurrence matrix, wherein each element in the matrix reflects the frequency of the simultaneous occurrence of gray levels of two pixel points in a specific distance and direction; S409, extracting useful texture features from the gray level co-occurrence matrix, and analyzing texture information in the image according to the extracted texture features; S410, detecting contours in the image by using a Hough transformation contour detection algorithm; s411, extracting coordinate information of the contour to form a contour curve; S412, matching the detected outline with a predefined standard shape template; s413, performing similarity calculation by using the shape descriptor, and determining whether the shape in the image matches a predefined standard shape according to the similarity result; s414, if the shape is successfully matched, recording and triggering a subsequent processing step, and if the shape is not matched or can not be identified, continuing to analyze other characteristics or images.
7. The method for monitoring the illegal behaviors based on double prevention according to claim 6 is characterized by classifying and identifying the extracted static behavior features and classifying and identifying the extracted dynamic behavior features, wherein the method comprises the following specific steps: S501, extracting the characteristics of the object, such as position, shape, size, color and the like, which do not change with time, in the video, analyzing the characteristics of the object, such as motion trail, speed, acceleration, direction change and the like, which change with time; s502, selecting a convolutional neural network CNN classification algorithm according to actual requirements and data characteristics; s503, training a classification model by using a data set containing known offence characteristics, so that the model can learn and identify different types of offence; S504, inputting the characteristics to be identified into a trained classification model, and outputting a prediction result, namely the specific type of the violation behavior by the model; s505, marking the recognized violation behaviors, including time, place and violation types; and S506, storing the identification result into a database.
8. The dual-prevention-based illegal behavior monitoring method according to claim 7, wherein the method is characterized by comprising the steps of carrying out risk assessment on the illegal behavior according to the nature and the result of the illegal behavior to determine the hazard level of the illegal behavior, outputting the identification result to related personnel in the form of images, videos or alarm information, and taking corresponding treatment measures aiming at the identified hidden danger, wherein the method comprises the following specific steps: S601, calculating a risk value for each identified behavior feature according to the risk level and the corresponding weight, wherein a risk value calculation formula is that the risk value = risk level multiplied by the weight; S602, summarizing risk values of all the identified behavior features to obtain an overall risk assessment result; s603, dividing risks into different grades according to the overall risk assessment result, and setting a clear threshold value for each risk grade; s604, selecting an evaluation result output form including an image, a video and alarm information according to the actual condition; s605, encoding the identification result and the risk assessment result; s606, formulating a template of output content, wherein the template comprises key information such as behavior feature description, risk assessment results, risk levels, occurrence positions, time stamps and the like; S607, sending the formulated output content to related personnel in a selected form; And S608, according to the risk assessment result, the related personnel select proper governance measures from the measure library to recommend.
9. The method for monitoring the behavior of violations based on double prevention according to claim 8, wherein for static behavior features and dynamic behavior features not identified as risk behaviors, predicting whether the behavior features are likely to generate risk after a certain period of time by a machine learning algorithm comprises the following specific steps: S701, screening static and dynamic characteristic data which are not recognized as risk behaviors by the existing monitoring system from a data source; s702, preprocessing the collected data, including removing repeated data, processing missing values and detecting and correcting abnormal values; S703, extracting possible features from the cleaned data, wherein the features comprise image features, time sequence features and statistical features; s704, evaluating the influence degree of each feature on the prediction result by using a correlation analysis method; S705, selecting a feature subset with obvious influence on a prediction result according to the feature importance evaluation result; S706, selecting a proper machine learning algorithm according to the problem characteristics and the data characteristics, selecting time sequence analysis for the data with time sequence characteristics, and selecting a random forest algorithm for processing multi-feature and high-dimensional data; s707, dividing the marked data into a training set, a verification set and a test set according to a certain proportion; S708, constructing a prediction model by using a selected machine learning algorithm, wherein the prediction model comprises model initialization and parameter setting; S709, training the model by using training set data, and adjusting model parameters by using a gradient descent iterative optimization algorithm so that the model can accurately predict potential risk behaviors; s710, performing performance evaluation on the trained model by using data of a verification set and a test set, wherein evaluation indexes comprise accuracy, recall, F1 score and area under ROC curve AUC; S711, carrying out deep analysis on the situation of model prediction errors, and finding out the reasons and possible improvement directions of the errors; s712, according to the performance evaluation result and error analysis, adjusting model parameters or improving model structures so as to improve the prediction performance of the model; s713, repeating the process of constructing, training and evaluating the model; And S714, inputting the real-time image data into the model to predict the risk behaviors.
10. The method for monitoring the illegal behaviors based on double prevention according to claim 9 is characterized by outputting a prediction result to related personnel and making risk prevention and control measures in advance for predicting possible risk behaviors, wherein the method comprises the following specific steps: S801, the behavior characteristics predicted by a machine learning model and possibly generating risks and relevant prediction results thereof are arranged, output data are formatted, the data are clear and easy to understand, and the requirements of different receivers are met; s802, determining specific related personnel to which a prediction result is transmitted according to an organization structure and responsibility allocation; S803, selecting a proper output mode according to the preference and the actual situation of the receiver; S804, sending the sorted prediction result to related personnel in a selected mode, confirming that the receiver receives the prediction result and knowing the content of the prediction result; s805, analyzing behavior characteristics and potential influences of the behavior characteristics possibly generating risks in the prediction result; S806, evaluating the severity, possibility and emergency degree of the risk, and determining the priority of the prevention and control measures; s807, specific prevention and control measures are formulated aiming at the predicted potential risk; s808, allocating corresponding manpower, material resources and financial resources according to the requirements of prevention and control measures; S809, implementing according to a preset prevention and control measure plan; S810, periodically monitoring the accuracy of the predicted result, and analyzing the difference and the reason between the predicted result and the actual occurrence; s811, evaluating the machine learning prediction model according to the monitoring result, and adjusting parameters and structure of the model according to the evaluation result so as to improve the prediction accuracy; s812, evaluating the implemented prevention and control measures, and analyzing the actual effect and the existing problems of the prevention and control measures; S813, establishing a continuous monitoring and iteration mechanism, continuously collecting new data, updating a prediction model and prevention and control measures, and continuously improving and perfecting the whole double prevention monitoring system.

Description

Dual-prevention-based illegal behavior monitoring equipment and method Technical Field The invention relates to the field of violation monitoring, in particular to a double-prevention-based violation monitoring device and a double-prevention-based violation monitoring method. Background The method for monitoring the illegal behaviors in the prior art mainly depends on a video detection technology to realize real-time monitoring and recording of the illegal behaviors, tracks the process of the illegal behaviors through analysis of continuous video images and photographs through analysis and control, but the method for monitoring the illegal behaviors in the prior art lacks predictive analysis and risk prevention and control mechanisms, so that the early warning capability and prevention and control effect of a monitoring system are limited, and most of the method adopts a passive response mode, namely the illegal behaviors can be discovered and recorded only after the illegal behaviors occur, and the problems of early warning and prevention cannot be early warned and prevented. Disclosure of Invention In order to overcome the problems that the prior art lacks predictive analysis and risk prevention and control mechanisms, so that the early warning capability and prevention and control effects of a monitoring system are limited, and a passive response mode is mostly adopted, namely, the illegal behaviors can be discovered and recorded only after the illegal behaviors occur, and early warning and prevention cannot be performed in advance. The technical scheme is that the dual-prevention-based illegal behavior monitoring equipment comprises a high-definition camera, a mounting frame and a central processing unit, wherein the mounting frame is arranged on one side of the high-definition camera, the central processing unit is arranged on the inner side of the high-definition camera, the mounting frame is used for mounting the high-definition camera at a key point position to be monitored, the high-definition camera is used for capturing high-definition images and video information, and the central processing unit is used for analyzing and identifying illegal behaviors in the videos and the images in real time by adopting image processing and machine learning algorithms. The violation behavior monitoring method based on double prevention is characterized by comprising the following steps of: s1, presetting behavior characteristics and risk grades, and carrying out risk grade division according to the categories of different behavior characteristics; s2, acquiring high-definition video image data in real time through a high-definition camera; s3, carrying out graying, noise reduction and enhancement pretreatment on video image data captured by a high-definition camera; s4, extracting characteristic information such as edges, textures, shapes and the like from the preprocessed image; s5, classifying and identifying the extracted static behavior characteristics; s6, classifying and identifying the extracted dynamic behavior characteristics; S7, carrying out risk assessment on the illegal behavior according to the nature and the result of the illegal behavior to determine the hazard level of the illegal behavior, outputting the identification result to related personnel in the form of images, videos or alarm information, and taking corresponding treatment measures aiming at the identified hidden danger; S8, predicting whether the behavior features of the static behavior features and the dynamic behavior features which are not recognized as risk behaviors are likely to generate risks after a period of time by a machine learning algorithm; And S9, outputting a prediction result to related personnel, and making risk prevention and control measures in advance for predicting possible risk behaviors. The method comprises the steps of presetting behavior characteristics and risk levels, carrying out risk level division according to different behavior characteristics, determining the behavior characteristics to be monitored, dividing the behavior characteristics into different risk levels according to the possible result severity degree of each behavior characteristic, and establishing a mapping relation between the behavior characteristics and the risk levels, wherein the specific steps are as follows: s101, setting a monitoring target and a monitoring range, namely determining which types of behavior features need to be monitored; S102, taking industry standards and specifications as the basis for defining behavior characteristics and classifying risk levels; s103, creating an exhaustive list to list all behavior feature categories to be monitored; s104, according to analysis of related laws and regulations, industry standards and historical accident cases, formulating standards of risk classification, wherein the standards relate to the severity degree, occurrence frequency and influence range of consequences possibly caus