Fusion method and device for adaptive video sensors

Abstract

Embodiments of the present invention propose an adaptive video sensor fusion method and device. The method includes: acquiring video information of a moving target captured by different video sensors, calculating the tracking accuracy of the moving target as the reliability of the different video sensors, and obtaining position information of the moving target tracked by different video sensors, said The position information includes the position of the moving target; the position of the moving target obtained from different sensors is mapped to the top view of the monitoring scene by using homography transformation; the position information of the moving target from different sensors is based on the different video sensors The credibility of the moving target is filtered by a local adaptive filter to obtain the optimized position and state estimation of the moving target, and the position and state estimation of the moving target optimized by different sensors are fused. The embodiments of the present invention have the feature of improving the fusion accuracy of video data from multiple video sensors.

Description

technical field [0001] The invention relates to the field of multi-sensor data fusion, in particular to an adaptive video sensor fusion method and device. Background technique [0002] With the development of multi-sensor systems and communication technologies, sensor fusion is applied at different levels and in different approaches. Signal-level fusion is the lowest level of fusion. In the field of video surveillance, it can be used to combine images acquired by homogeneous or heterogeneous sensors in the same scene. Before actually performing the fusion method, the images to be fused need to be correctly spatially registered. Signal-level fusion is the most restrictive, as it can only fuse image / video-like sensors that output images or video. [0003] Compared with the limitations of signal-level fusion, feature-level fusion allows the video sensors to be fused to have different fields of view, such as different viewing angles, different distances from the target, etc., ...

AI Technical Summary

Problems solved by technology

The above fusion methods are rarely applied to the actual video sensor fusion, and the reliability of the input sensor is not considered, so that a faulty sensor may affect the fusion process, resulting in inaccurate fusion results

However, the existing sensor reliability function can not meet the requirements of most practical video surveillance systems, such as real-time, easy to implement, etc.

Images