基于PCNN的高斯混合模型运动检测改进方法

Improved Gaussian mixture model for moving detection method based on PCNN

针对固定摄像机的视频监控系统,提出了一种改进的基于混合高斯模型的运动目标检测方法.改进方法引入PCNN算法,针对模型匹配问题,提出自适应局部阈值算法并结合区域增长思想,利用PCNN的迭代计算,逐步检测出运动目标.实验表明,改进的方法与传统方法相比具有更好的运动目标检测能力,在运动目标和背景的灰度值差别比较小的情况下,能改善其运动目标检测的效果.

An improved moving objects detection method was proposed in the paper based on Gaussian mixture model in the case of focusing on a video monitoring system with a static camera. Compared with well-known algorithms, the proposed method had the following two features. First, for matching the existing Gaussian distributions, the adaptive threshold based on neighborhood was attained by PCNN. Second, through circuit calculations of PCNN, moving objects were detected step by step, and the region growing algorithm was used in the procedure. Experiment results show that the proposed solution possesses a better ability to detect an object when it is in a low contract with the background.