Internationale Zeitschrift für Fortschritte in der Technologie
Offener Zugang

Abstrakt

Object Detection Algorithm in Underground Mine Based on Sparse Representation

Yan Lu and Qing-xiang Huang

This paper puts forward an improved K-SVD object detection algorithm for the problem of multiple noise sources in underground mine video. Firstly, the background modeling is applied in the video; then, the improved non-local mean filtering algorithm is used to enhance the image quality; finally, the improved image is processed by the sparse representation algorithm to further detect the moving object. In order to verify the effectiveness of the proposed algorithm, the algorithm and other algorithms are applied to video object detection in two different scenarios. The experimental results show that, in the underground mine video, the proposed algorithm can increase the accuracy by more than 8% compared with the traditional K-SVD algorithm, and the proportion of error points decreases by about 25%. Better detection of the moving object is achieved by the proposed algorithm.