Multi-target tracking method based on multi-characteristic binding in combination with Camshift algorithm

Abstract

The invention discloses a multi-target tracking method based on multi-characteristic binding in combination with a Camshift and relates to the field of computer visions. The method comprises three steps of S1, performing fusion detection on multiple moving targets based on a background subtraction method and a frame difference method, wherein the detection of the multiple moving targets includes three steps: (1) initializing a background model; (2) updating a background by use of the frame difference method, then performing binaryzation, and (3) performing background subtraction by use of thebackground subtraction method; S2, further denoising the binary image; and S3, performing tracking of the multiple moving targets based on a Kalman filter and a Camshift algorithm. According to the method, the multiple moving target are detected in fusion by use of the background subtraction method and the frame difference method, meanwhile, the exact moving targets are extracted, most of noise ina moving detection process is eliminated by use of repeated morphological operations, the situation of deformation of the target can be adapted in the target tracking process, correct tracking can beperformed when the target is shielded, therefore, the accuracy of intelligent video monitoring is greatly improved, the monitoring stability is enhanced so that the real-time demand is satisfied.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a multi-target tracking method based on the combination of multi-feature combination and Camshift algorithm. Background technique [0002] With the continuous development and deepening of various policies such as my country's safe city construction, and the continuous enhancement of security awareness of users in various industries such as transportation, education, and finance, the video surveillance market has grown strongly, the number of cameras has increased rapidly, and video resources have exploded. However, the monitoring method is still mainly manual monitoring, which brings many problems, such as fatigue of monitoring personnel, many false positives and missed negatives, difficult video retrieval, a large amount of garbage data, etc., and the real-time performance of the video monitoring system cannot be effectively utilized. [0003] In order to solve the above problems,...

AI Technical Summary

Problems solved by technology

However, when the target color is similar to the background color or the lighting conditions change, the accuracy of tracking the target will be greatly affected, and it is easy to fall into the local maximum when the moving target is occluded. fail

Images