为解决现有视频监控系统中目标检测算法无法应付复杂的环境且计算量较大等问题,结合背景模型算法和帧间差分的优点,对混合高斯背景方法和帧间差分进行改进,提出一种基于混合高斯模型背景法和混合差分相结合的运动目标检测改进算法.利用分块思想进行高斯背景建模,利用多帧差分实现混合差分,既能得到较高的灵敏度又能进一步提高检测效果和速度.通过实验证明该算法的可靠性和实时性.%In order to solve the problems such as the complex environment that the object detection algorithm for available video surveillance system cannot deal with and the greater amount of computation, the Gaussian mixture background method and frame difference method were improved by making use of the merits of background model algorithm and frame difference method. An improved moving objects detection algorithm was proposed based on Gaussian mixture model background method combined with hybrid difference. A blocking idea was adopted to perform Gauss background modeling and the multi-frame difference was used to realize the hybrid difference. By doing so not only a higher sensitivity could be obtained but the further improvement on detection speed and effect could also be achieved. It was verified by experiment that the real-time ability and reliability of this algorithm.
展开▼