Method for tracking multiple moving objects based on combination of disparity map background subtraction method and Meanshift

Abstract

The invention discloses a method for tracking multiple moving objects based on the combination of a disparity map background subtraction method and Meanshift, and relates to the field of computer vision. The method mainly comprises the three steps of S1, detecting moving objects and removing the ghost based on Vibe and the disparity map background subtraction method; S2, carrying out a multi-object segmentation algorithm; and S3, tracking multiple moving objects based on multi-feature combination and a Mean Shift algorithm. The method has the advantages that 1) the integral moving objects canbe extracted, a problem that monocular vision based moving object detection is susceptible to the influences of illumination and shadow, and the ghost phenomenon can be eliminated at the same time; 2)all moving objects can be separated and recorded; and 3) the moving objects can be accurately tracked on the basis of a tracked object feature extraction technology, the effect is good, and the operation speed is high.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a moving target detection method based on the combination of disparity map background difference method and Meanshift. Background technique [0002] Moving target detection is the basis of target recognition and tracking. Fast and accurate detection of moving targets is conducive to follow-up work such as target tracking, recognition and behavior understanding. It has a wide range of applications in iris, face recognition, security monitoring, robot navigation, aircraft and satellite monitoring systems, etc. [0003] Among them, there are three kinds of moving target detection algorithms: optical flow method, frame difference method and background difference method. The optical flow method requires special hardware support, and the calculation is complex and computationally intensive, so it is generally seldom used. The principle of the inter-frame difference method is simple, an...

AI Technical Summary

Problems solved by technology

The optical flow method requires special hardware support, and the calculation is complex and computationally intensive, so it is generally less used

The principle of the inter-frame difference method is simple and insensitive to noise and light changes, but the target detection results are prone to holes

The background method can extract the complete information of the target, but it is easily affected by the dynamic changes of the external scene such as lighting.

When the moving object changes slowly to fast motion, the background subtraction method can easily detect the background exposed area (that is, the current background still has the moving object information of the previous frame, but the moving object is no longer in this area at this time) as the foreground , the phenomenon of "shadow" appears, and the noise processing effect is not good in complex scenes such as tree branches swinging, and the adaptability to the environment is poor. Such false detections will cause difficulties in subsequent target tracking.

The traditional moving target detection method based on monocular vision can detect the outline of the moving target, but it is easily affected by external conditions, and will detect shadows and part of the background as foreground

Images