一种基于优化ViBe算法的人形分割方法

A Humanoid Segmentation Method Based on Optimized ViBe Algorithm

人形分割在视频监控研究中具有极其重要的意义。ViBe算法因其运算效率高、算法思想简单且容易实现的特性受到许多研究学者的青睐,并被应用到人形分割的研究中。针对ViBe算法在应用于人形分割研究时易于出现鬼影和容易对光照引起的阴影造成误检的现象,提出了一种改进的ViBe算法—VSFC(ViBeSurendraFive-frameCanny)算法。首先将利用Otsu自适应阈值的Surendra算法得到的初始背景用作背景模型的初始化。其次,通过将ViBe算法所得出的前景与五帧差分法得出的前景采用“与”运算的方式获得人形分割结果,并对得到的人形分割结果进行形态学处理,以便能够消除较大的噪声。最后利用Canny边缘检测算子对人形的边缘信息进行提取,利用小波融合方法将前面得到的人形结果和边缘信息进行融合,得到最后的人形分割结果。在更新机制方面,提出动则更新的更新思想,对于那些长时间静止不动的像素点,采取隔时更新的方式,提高了算法的实时性。

Humanoid segmentation is extremely important in video surveillance research.The ViBe algorithm is favored by many researchers because of its high computational efficiency, simple algorithm and easy to implement, and is applied to the research of human segmentation.Aiming at the phenomenon that ViBe algorithm is prone to ghosting and easy to cause false detection of shadow caused by illumination when applied to human-shaped segmentation, an improved ViBe algorithm-VSFC (ViBe Surendra Five-frame Canny) algorithm is proposed.The initial background obtained using the Otsu adaptive threshold Surendra algorithm is first used as the initialization of the background model.First, the initial background is obtained using the Otsu adaptive threshold Surendra algorithm, and this background is used as the initialization of the background model.Secondly, the human-shaped result is obtained by using the foreground obtained by the ViBe algorithm and the foreground obtained by the five-frame difference method to obtain the humanoid result, and the obtained humanoid result is processed morphologically to eliminate the large noise.Finally, the Canny edge detection operator is used to extract the edge information of the humanoid.The wavelet fusion method is used to fuse the previously obtained humanoid result and edge information to obtain the final humanoid segmentation result.In terms of the update mechanism, the update idea of the update is proposed.For those pixels that are still stationary for a long time, the method of updating at intervals is adopted to improve the real-time performance of the algorithm.