Research On Multi-channel Visual Perception System Technology For Smart Construction Site

With the increasing progress of society and the rapid development of technology,the requirements for intelligent and intelligent degree of safety risk supervision of construction sites are getting higher and higher.At present,most surveillance cameras on construction sites work based on a single algorithm and have relatively single functions;some cameras may be equipped with multiple algorithms to adapt to different locations or scenarios,but the settings need to be readjusted during or after the movement,which is inconvenient to use;and There is a lack of linkage between multi-channel cameras,and automatic division of labor is impossible,requiring manual adjustment by personnel;if the video content is completely detected,the computing power requirement is high and the real-time performance is low.Therefore,the automatic detection rate of construction site video is low,and a large number of videos cannot be detected in time,resulting in poor flexibility in the intelligent analysis process of the construction site monitoring system,it is difficult to dynamically define the perception process,and it is difficult to form a dynamic multi-algorithm combined detection in the whole process and in the whole time domain.It even causes a lot of waste of video content.In response to the above problems,this paper designs a flexible real-time analysis framework FRTA for surveillance video;designs a dynamic definition perception process,and forms a multi-channel visual perception scheme for intelligent construction sites with dynamic multi-algorithm combined detection in the whole process and in the whole time domain;The multi-channel visual perception system for smart construction sites solves many problems in the process of system design and implementation.The main work includes:(1)The flexible real-time analysis framework FRTA for surveillance video is designed:the algorithm and algorithm version are dynamically selected according to various factors such as system resource usage 、 task urgency and importance,perception algorithm accuracy,perception algorithm efficiency,etc,which solves the problem of single camera function and video content.Unable to be detected in time and camera movement monitoring problems;the combination of task programming and blackboard mode can efficiently generate video stream tasks,solve data multiplexing and ensure the timing effectiveness of video frame sequences,and realize multi-intelligent algorithms to detect and analyze ordered videos at the same time unordered calculation of frames;priority queues and time-limited waiting are designed through the unique identification of video frames and the small root heap algorithm to ensure the orderly and real-time output of video frames after processing.(2)The dynamic definition perception process and multi-channel visual perception scheme are designed: through the design of fast focus adaptation algorithm,event classification,setting video frame memory existence time,etc,high concurrency technology is used to calculate in the order of task generation to avoid the existence of a large number of pictures causes memory overflow to ensure the fluency and real-time performance of the video;through source stream multiplexing and task dynamic binding to connect multi-channel cameras,realize automatic allocation of algorithm levels for multi-channel video streams,achieve mutual cooperation,avoid repeated detection,or focus on improving detection Accuracy;separate the detection results from the original video stream to ensure that the video source stream is not contaminated;draw the detection and analysis results in the video frame,and reflect the detection and analysis results in the video stream.(3)Design and implement a multi-channel visual perception system for smart construction sites: design interface adapters compatible with common market cameras and other equipment;realize user-defined video streams and intelligent algorithm combinations;set detection levels according to user needs,flexibly allocate resources,and avoid resources Waste,reduce system construction cost;adopt intelligent and flexible design,make the system easy to expand and maintain,and improve the level of construction site intelligence.