Moving shadow detection is critical for video surveillance system since shadow points are often misclassified as object points,which causes errors in segmentation and tracking.The conventional algorithms always need the scene characteristics,which makes the algorithms not widely used and automatically performed.An approach is presented that does not rely on any scene assumptions such as camera location and scene characteristics.Color information,texture information and spatial constraints are embedded to define the overall algorithm.Firstly,relevant areas are identified in each image.Then,the color distortion caused by shadow are calculated. Finally,the moving object and shadow are detected by using color distortion compensation and texture verification.The proposed algorithm is demonstrated on many kinds of outdoor video sequences.Moreover,the performance comparisons show that the proposed algorithm outperforms the conventional ones.

Block matching algorithm is playing an important role in motion estimation at present. Multi-level successive elimination algorithm(MSEA)is an exhausted block matching motion estimation search algorithm, which has a precision character. But its operation speed can not meet the requirement of real time application only by software method. The paper tries to achieve the real time using FPGA design. Firstly, it has been designed with basic FPGA hardware, and then the pipeline structure is applied to the algorithm. The speed of basic hardware method is over 14 times than that of the optimized software method. The speed is higher when the pipeline structure has been introduced into the basic hardware method. So the real time monitoring of moving target becomes possible.

It's a hot field about motion estimation now,which has an abroad application in video coding and target real time tracking.It's common to use block matching algorithm(BMA)at present.The classic algorithms of BMA are SEA and MSEA.The paper has improved them at two sides at the base of analyzing the algorithms.Calculation of the improved algorithm has decreased effectively from the experiment result and the veracity has no change.

Traditional two-point algorithm and two-point multi-section algorithm are the linearity corrected calibration-based techniques.Two-point algorithm needs less operation time but has lower precision.Traditional two-point multi-section algorithm needs much more operation time for getting higher precision.Aimed at the limitation of two-point multi-section algorithm whose precision is dominated by the number of section,we make some improvement on it.This improved two-point multi-section algorithm uses the auto-split-section method,so needs smaller space to store parameters and less operation time.After this improvement,we mix two-point algorithm and improved two-point multi-section algorithm into a mixed algorithm of two-point multi-section.This mixed algorithm uses these two algorithms for different optical sensor.These designs not only enhance the real-time characteristic but also improve the precision.

This paper presents an Automatic Vehicle Classification System for random moving vehicles. In training step, we obtained 360 projections of each vehicle by rotating video camera round it at same observation angle. The rolling, transferring and scaling (RTS) invariant features of each projection have been achieved. It made a RTS invariant curve of each target. In classification step, the motions of each target were detected in the video with the same observation angle. The RTS invariant features have been achieved. And the target were classified by calculating the distance between motions'RTS invariant feature and RTS invariant curves. The experiment results show that this method is quite promising.

运动阴影常被误划为目标造成错误的目标分割和跟踪，所以阴影检测在许多图像监控系统中非常重要。本文研究了传统阴影检测算法，针对传统算法受特定条件约束不能自动适用于不同场景的不足。提出了一种针对摄像位置和视频特征未知情况下的阴影检测算法。算法综合考虑了颜色信息、空间信息和纹理信息，利用阴影的颜色、空间和纹理属性在待分析区域中确定其造成的颜色形变，通过使用颜色形变补偿和纹理校正检测到了运动阴影。基于不同图像的实验结果表明了该方法的有效性。