基于特征信息提取的目标识别算法研究-通信与信息系统专业论文.docx

ABS1RACT ABSTRACT Object reω伊ition is currently one of the most active research topics inωmputer vision. Aft町 decades of research and development， object reco伊ition technique has been developing fast and its wide use is improving the quality of peoples life consid叮ably in various fields. Nowadays ，urban transport facilities have been improved and 也e increasing number of vehicle leads to an increasing accident rate，so inte1ligent traffic surveillance is imperative. Moreover，位affic object reω伊ition is 也e groundwork of the intel1igent traffic surveil1ance system. Further research need to be done to improve the object rec。但?tion algorithm since it has a significant inf1uence on traffic surveil1ance. Consid配ing the characteristics of 加 traffic object，伽 feature based object reω伊ition a1gorithm hωbeen studied. B础。d on these studies，SVM b础。d object reω伊?tion a1gori也m and object rec。但tion a1gorithms based on keypoint matching are presented. Due to the ∞mplexity of the traffic scene and the particularity of the moving obje仗， some kinds of fea阳re extraction me由ods fit for 仕affic object have been anal归叫.A仕即 in-d叩，th study of support vector machine 由eory and its application principles ，SVM bas叫 object recognition a1gori也m is proposed ，constructing a SVM based multimoda1 classifi町 andthen 也e traffic object can be re∞gnized. Exp创menta1 results show 由at 也is method is effective. In addition，the ωmparison of the recognition accurate for different feature extraction methods provides a basis for feature selection. Furthermore，an object re∞伊ition a1gori也m based Kalman mul也object tracking a1gorithm is presented. The object recognition a1gori也m resolves the issue of updating 也ep町皿net町s of Ka1man filter. Experiment a1 results show 也at these modifications fur由町 improve tracking p町formance. However，the results of SVM based object reω伊ítion a1gorithm demons位爬出at 伽e effect of object reωgnition algori由m wi11 be affected by the object detection result. In order to resolve this problem