Method and system for feedback-type object detection and tracing of video object

Abstract

The invention discloses a method and a system for the feedback-type object detection and tracing of a video object, relates to the field of image processing and solves the problems that a hole occurs inside a foreground moving object in a current frame and that the prior proposal fails to obtain higher-quality boundaries. The method comprises the following steps: offering object detection to the video image newly acquired by each frame to extract the foreground moving object of the current frame; dividing blocks according to a foreground moving object of a video image of the former frame (a reference frame) to obtain a reference block; using a mean shift algorithm to trace the reference frame to calculate a candidate block of the foreground moving object in the current frame; and then determining the final foreground moving object according to a rate of superposition area of a tracing result and a detection result and feeding back a result to update a ground model.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method and system for feedback detection and tracking of video objects. Background technique [0002] In image processing, when analyzing a moving object of interest in a video stream, it is necessary to detect and track the moving object. In the prior art, the detection and tracking of the moving object are two independent processes, the object detection is performed on the image first, and then the detection result is tracked. [0003] The most common method for object detection is background subtraction. The background elimination method first uses the first few frames of the video stream to establish a background model, and then divides each subsequent frame of video images into foreground moving objects and background according to the background model. Compare the pixel area of ​​each subsequent frame of video image with the corresponding pixel area of ​​the background mo...

AI Technical Summary

Problems solved by technology

[0006] When performing object detection, when the moving object is large in scale or moving slowly, because the edge part of the moving object is similar to the background, it may not be possible to clearly separate the outline of the edge part from the background, causing false detection results. thus creating a void

[0007] The object tracking method can track the center or approximate range of the foreground moving object, but cannot correctly match multiple foreground moving objects with a high degree of overlap or overlap, resulting in a "mismatch" phenomenon, resulting in tracking failure

Images