| In recent years,with the development of society and economy,more and more people pay attention to the security of public places.Carrying object detection in video surveillance is one of the hot spots in the research of intelligent video surveillance,It has important applications in the areas of security,public security and so on.On the basis of existing algorithms for video surveillance was carried pedestrian detection were studied,we present a carried object detection method based on frequency matching,in the actual scene surveillance video images on the test results show that there is good detection results.Specific research work the following aspects:(1)aiming at the existing current moving target detection algorithm is commonly used for research,and analyzing the advantages and disadvantages of these algorithms.Choose the Codebook algorithms as the goal of this article segmentation algorithm,The detection results of each algorithm are compared by experiments.Finally,use the method of mathematical morphology to preprocessing the target image;(2)Analysis the contour features of human body in the condition that carrying objects and without carrying objects,the results show that the human body contour is symmetrical about body axis,when the human body without carrying objects.and the body of the nonsymmetric regions of the central,axis extract the body with physical state.This part that the preliminary asymmetric regions is judged.(3)On the basis of persons with early judgment was based on the analysis of the period features interference region presented by frequency matching method in this part of the region has been ruled out,got the final position of the items,and by law section of the article species were judged.The algorithm is recommended for the actual scene video image experiments to obtain good detection results. |
