基于RGB三通道分离的运动目标检测方法

摘要

Extraction and update of the background is the key to background subtraction. The traditional background subtraction is processing gray image, while lots of information will be lost after converting color images to gray images. In this paper, the time median method is improved, and an moving object detection method based on RGB three - channel separation is designed and realized. After that, we deal with the binary image by morphological and connectivity detection to elimitate the noise. Finally, the shadow part is removed in HSV space. The experimental results show that the algorithm can extract background model even in the presence of moving object, and also can detect the object when the gray of target is similar to the gray of background.%背景图像的提取和更新是背景差分的关键.传统的背景差分法是对灰度图像进行处理,在检测前要对彩色图像进行颜色空间的转换,必然会丢失许多信息.对时间中值获取背景模型的不足加以改进,设计并实现了一种基于RGB三通道分离的运动目标检测方法.用形态学处理和连通性分析消除噪声,用区域填充技术填充目标区域内部空洞,在HSV空间去除阴影部分,得到比较准确的运动目标.实验结果表明,该算法在运动目标存在的情况下也能获得较准确的背景模型,当目标灰度值和背景灰度相近的时候,也可以检测到较完整的运动对象.