中文摘要：

基于减背景技术提出了一种改进的运动目标检测模型CW4,与原模型W4相比,CW4充分利用了图像的亮度、色度和饱和度等颜色信息,使得目标检测的结果更加准确。在对带有阴影的彩色图像的背景和前景的像素特点进行分析后,还设计了一种带权重的颜色计算模型的阴影去除算法。实验结果表明,基于CW4的算法显著提高了行人检测的精确性,阴影去除算法也能够有效地检测和去除阴影。

英文摘要：

Pedestrian detection and behavior recognition have become major research topics in intelligent video surveillance. A typical system of pedestrian detection and recognition includes modules like pedestrian detection, shadow removal, pedestrian tracking and behavior recognition, in which object detection is the basis of pedestrian detection and behavior recognition technologies. Based on the background subtraction thchniques, an algorithm named CW4 is proposed in this paper. As the improved version of W4,it makes full use of the color information such as hue, saturation and intensity of images,and makes the detection results more accurate. After analyzing the pixel features of background and objects with shadow, a computing model which assigns different weights to three components of HSI information is designed for shadow removal. The experimental results show that the algorithm CW4 can significantly improve the accuracy of pedestrian detection, and the shadow removal algorithm can remove the shadow efficiently.