Moving Object Detection Based On Linear Clustering And Difference

A moving object detection technique is a process of obtaining a moving object from a continuous sequence of video pictures taken by a camera. So far, the moving target detection technology has been widely used in indoor and outdoor real-time monitoring,security and other fields. On the basis of predecessors’ theory,there are still some difficulties in the direction of moving target detection technology, such as shadow and noise. The research algorithms mainly include background modeling,feature extraction .and super pixel segmentation. The main work of this paper is as follows:Aiming at the problem of using the mixed Gaussian model to construct the background and obtaining the moving object by the background difference,it can’t extract the accurate result and noise due to the influence of the shadow. A moving object detection method is proposed. Firstly, the background image is constructed by using mixed Gaussian. Secondly, a simple linear iterative clustering is used to locate the target frame image, and the background is separated from the target to avoid the interference of the shadow. The localized three-valued pattern texture is extracted from the segmented frame image and the background image Feature, to reduce the impact of noise in the inter-frame difference on the extraction results, and finally compare the similarity between the frame to be detected and the background frame to obtain the final experimental results. The experimental results show that the algorithm can obtain high precision and recall rate, and obtain a more complete target image.A new method of moving target detection is proposed, such as ghosting, target area, and so on. First, the current frame and the adjacent frame of the current frame are divided; secondly, the threshold is set using the Otsu threshold segmentation method, and the current frame and the adjacent frame are subjected to the difference operation to obtain the approximate region where the moving object is located; The way in which the current frame is divided is the set of regions containing the target.Finally, by comparing the difference results with the segmented result graph, the target is obtained. The results obtained by the experiment show that the algorithm can obtain high precision and recall rate and improve the integrity of target detection.