A Long-term Stationary Object Detection and Tracking Method with State Annotation

Abstract

A long-term stationary object detection and tracking method with state labeling. Firstly, a long-short dual foreground model is constructed, and a foreground mask is generated after fusion. Then, the blobs are classified and marked based on the motion state of the blobs on the foreground mask, and the long-term blobs are obtained. block, and finally update the long and short double background models of the corresponding regions of the different states of the clumping; Classify and mark the movement state of the clumps based on the duration time difference of the clumps, the movement trajectory of the clumps, and the difference between the clumps and the surrounding pixels, and use the logistic regression analysis method to obtain the probability value of the movement state of the clumps. The long-short dual-background model update of the corresponding regions of different state blobs, the detection process is real-time, and the video detection effect and robustness for different scenes are better than other current methods, which can meet the needs of practical applications.

Description

technical field [0001] The invention belongs to the field of computer vision object detection and tracking, and in particular relates to a long-term stationary object detection and tracking method with state marking. Background technique [0002] Object detection and tracking based on video has important applications in the fields of video processing, computer vision and pattern recognition. Illegal parking, a traffic law problem, has become increasingly prominent. Illegal parking has increased the incidence of traffic accidents and reduced traffic efficiency; burden and cannot guarantee the timeliness and objectivity of law enforcement. In contrast, if intelligent video monitoring and detection of long-term parked vehicles in traffic videos can be carried out, the purpose of illegal parking supervision can be achieved, which can not only reduce labor costs but also improve efficiency. Therefore, it is necessary to propose an accurate and fast long-term stationary object d...

AI Technical Summary

Problems solved by technology

[0021] In order to solve the problems existing in the actual application of existing long-term stationary object detection and tracking methods, the purpose of the present invention is to provide a long-term stationary object detection and tracking method with state annotation, which can quickly, accurately and robustly detect various Long-term Stationary Objects in Videos of Complex Scenes

Images