A New Approach Based on Multi-feature for Cooperative Target Detection

Abstract

Cooperative target has been widely applied in vision-based navigation for unmanned aerial vehicle. For overcoming the targets susceptibility of the surroundings, a fast and accurate detection approach based on multiple features is put forward. Firstly, utilize the pyramid image preprocessing method to eliminate some noise. Then the image features consisting of image contours, Hu moment invariants and FAST corners, are selected to judge and identify the cooperative object. Moreover, the algorithm is improved and accelerated in order to meet the real-time requirement. Experiments show that the proposed approach has a better adaptability and robustness for cooperation target recognition, under different scales, different angles and environment disturbance.