Research on moving human and vehicle's recognition algorithm

Abstract

SIFT algorithm and optical flow algorithm have been used. Combining the merit of SIFT algorithm that it can match object accurately under the situation of image scale and rotation change, illumination change, mask and deformation with the Characteristic of optical flow algorithm that it can extract moving object well, aiming at the moving person and vehicle's detection and recognition, the article proposed new method which is the optical flow algorithm uniting the SIFT algorithm. This method only needs that the template goal is similar to the actual target, then we can judge the goal's category accurately. It is unnecessary that the match must meet the condition that the identical person and the identical person template or the identical vehicle and the identical vehicle template to match, then determine the goal. Therefore the method is simple and convenient. The experimental results indicated that the algorithm's accuracy is high, its operating speed is quick, which can meet the needs of actual detection.