Multiple angle movement target detection, positioning and aligning method

Abstract

The invention discloses a multi-view moving target detection, positioning and corresponding method, which belongs to the technical field of video monitoring. The method includes the following steps: performing foreground detection on video images of multiple viewing angles to obtain a binary foreground image; establishing a space field model according to the binary foreground image, and performing three-dimensional reconstruction in the space field model to obtain a motion The 3D reconstruction result of the target; analyze the 3D reconstruction result, detect and locate the moving target in the space field, and obtain the spatial position of the moving target; project to each viewing angle according to the spatial position of the moving target, and determine multiple viewing angles Correspondence between moving targets. The invention has the characteristics of strong ability to deal with occlusion, fast calculation speed and the like, and can meet the real-time requirements of video surveillance.

Description

technical field [0001] The invention relates to the technical field of video surveillance, in particular to a method for detecting, locating and corresponding a multi-view moving target. Background technique [0002] In recent years, video surveillance has been widely used in urban security, traffic monitoring, production safety and other fields. In video surveillance, the technical problems of multi-camera surveillance have been plagued by the development of the industry. Multi-camera monitoring includes two situations: one is that there is no overlapping area between the camera views, that is, the single-view situation; the other is that there are overlapping areas between the camera views, that is, the multi-view situation. [0003] The detection, positioning and corresponding technology of moving targets in multi-view situations is the key technology in the multi-camera video surveillance system, and is the basis for subsequent advanced processing such as target trackin...

AI Technical Summary

Problems solved by technology

When there are multiple moving objects, the problem of detecting, localizing and corresponding objects becomes complicated

There are mainly two problems in the following aspects: First, multiple targets interact with each other to generate occlusion, making it very difficult to accurately detect, locate and correspond to targets; second, the computational complexity of multiple viewing angles and multiple targets is high

The method based on the fusion space does not directly extract the target on the two-dimensional image of each viewing angle, but processes the space after information fusion. The fused information takes into account the influence of occlusion between targets, so this kind of method has better ability to deal with occlusion. Strong, often get better processing results, but the disadvantage is that the computational complexity is very high, unable to meet real-time requirements

[0006] In summary, the existing methods have weak ability to deal with occlusion, high computational complexity, and it is difficult to meet the requirements of video surveillance

Images