| With the rapid development of artificial intelligence,moving target detection has become a research hotspot.It plays a very significant role in target tracking and anomaly detection.After years of research,good results have been achieved by moving target detection in simple scenes,but in some complex scene the results of motion target detection is not ideal,such as illumination changes and large displacement interfering detection results in static background and camera jitter in the shooting affecting moving target detection results in dynamic background.In view of the above two problems,in-depth study was carried out.On the basis of the existing research results,a general moving target detection algorithm is proposed in static background,and an improved algorithm is proposed in dynamic background.First of all,the knowledge of image processing is introduced briefly in this paper.By studying three commonly used algorithms of moving object detection,their advantages,disadvantages and application scenarios are analyzed.Carry out experiments to verify them in this paper.Secondly,in view of the influence of illumination change and large displacement on moving target detection in static background,a general algorithm in static background is proposed in this paper.Image texture decomposition and Pyramid technology are combined to Lucas-Kanade optical flow algorithm.The experimental results show that this method has better detection results for the moving target detection in the static scene.Finally,aiming at the interference caused by camera jitter in dynamic background,a block matching based eliminate jitter algorithm is proposed.The optical flow field is calculated by the block matching of the image,and then the motion compensation is carried out to eliminate the global motion caused by the jitter ofcamera.Through the experiment,it is proved that this method can effectively eliminate global motion in dynamic background,and make the accuracy of detection higher. |
PDF Full Text Request