Based on Harris corner of Drones vehicle target image matching method

For acquiring vehicle top feature data what algorithm is used to study and its matching effect and other issues. Based on this, key recognition features such as vehicle top geometry features, especially sunroof, cab roof light and bus rearview mirror, are fully utilized to improve the UAV image target matching accuracy. In this paper, I use UAV vertical observation true-color image data, based on Harris and SIFT feature descriptors combined with the NNDR matching method, focusing on corner point feature extraction of top features of vehicles in images, and determine thresholds to match different types of known samples with images to be matched. The experimental data results show that the top lights of cabs are rich in corner point features, and their matching accuracy can reach more than 85%, which is higher than that of small cars with relatively few corner point features, both corner points are particularly critical features in the matching process. The method significantly improves matching for vehicle types rich in corner point features, and the top features make more use of geometric feature information, which is especially suitable for image data with higher spatial resolution and better image grayscale contrast.