Traffic monitoring method and traffic monitoring system

Abstract

The invention relates to a traffic monitoring method and a traffic monitoring system. The method comprises the following steps: S1, collecting video information of a road through a camera; S2, performing multi-target detection and identification on the video information acquired in the step S1 to detect vehicles and pedestrians in the video; S3, performing tracking and trajectory analysis on the multiple targets detected and identified in the step S2; and S4, according to the multi-target tracking and trajectory analysis result in the step S3, obtaining the speed and acceleration of the target, judging whether a traffic collision accident occurs or not, and if the traffic collision accident is judged to occur, automatically carrying out alarm processing. Video collection is standardized, and video sources can be effectively managed; automatic analysis of video structuralization is established, and post-event retrieval depending on manpower is liberated; and compared with an existing traditional target detection algorithm, the traffic monitoring method and the traffic monitoring system are high in detection precision and good in effect; and multi-stream video structural analysis canbe input at the same time, and results can be output.

Description

technical field [0001] The invention relates to the field of traffic monitoring, in particular to a traffic monitoring method and a traffic monitoring system. Background technique [0002] At present, the scale and scope of video surveillance systems continue to expand, and smart city video surveillance systems have entered the era of networked surveillance. Video structuring and big data applications based on high-definition surveillance are an important direction for the development and construction of smart cities in the future. However, the current video surveillance system has many defects. For example, in the face of video information, it is time-consuming, labor-intensive and inefficient to manually retrieve unstructured data after the incident, which makes the surveillance system prevent and control public security and investigate cases. The effect is greatly reduced; the existing traditional target detection algorithms are mainly based on HOG or DPM algorithms, whic...

AI Technical Summary

Problems solved by technology

However, the current video surveillance system has many defects. For example, in the face of video information, it is time-consuming, labor-intensive and inefficient to manually retrieve unstructured data after the incident, which makes the surveillance system prevent and control public security and investigate cases. The effect is greatly reduced; the existing traditional target detection algorithms are mainly based on HOG or DPM algorithms, which have the problems of slow detection speed and low accuracy, and it is difficult to achieve high enough matching accuracy; the monitoring system will Simultaneous access to multiple video streams, the current traditional target detection and recognition algorithm is still difficult to process multiple streams of video input at the same time

Images