| Intelligent video surveillance (IVS) technology is an emerging research orientation in the field of computer vision. Its main goal is to realize the description, understanding and analysis of the content of the surveillance video by integrating computer vision technology, image video processing technology and AI technology.Moving target detection and tracking is the most basic core technologies of IVS, which not only are the cornerstone of the following advanced processing and recognition, target classification, behavior analysis and recognition and semantic index and so on, but also the key for automation, intelligent and real-time application of a video monitoring system.This paper is committed to the key issues of IVS technology, which are target detection and target tracking. And the research work mainly covers the following topics:(1) Characteristics of Gaussian and pepper-salt noise are discussed. And the denoising methods for each noise are introduced. Finally a new denoising method, combining the median algorithm and mean method, is proposed to deal with the mixed noise.(2) In the area of the detection of moving objects, firstly, an improved background modeling method is proposed to obtain more efficiency and more stable model than traditional mean value method. A regional labeling method based on run-length-encoding (RLE) is used to improve the efficiency of traditional scanning method. And then the gestalt visual principles are introduced to integrate and optimize the detected objects. Experiments based on traffic video images are performed and good results are obtained.(3) In the area of object tracking, the traditional moving target tracking methods are introduced firstly, then a new method based on multi feature fusion matching is presented. The tracking result of this method is much better than using a single feature. Secondly, cubic spline interpolation is used to fit the tracking trajectory of moving objects. Experiments based on traffic video images are performed and good results are obtained. |
