结合形态学和假设检验的视频对象分割

视频对象分割是当前图像和视频处理的热点和难点之一。文章首先采用形态算子和改进的watershed算法对图像序列进行空间分割,然后利用F检测算法进行帧间变化检测,将时空分割结果结合起来,得到初始的变化检测模板。通过相应的基于二值形态算子的后处理,得到最终的分割结果。整个过程基本是对灰度图像和二值模板的形态处理,简单易行。实验结果表明该算法可以较好地分离前景和背景,定位和分割视频对象。

The Video Object Segmentation Based on Combining Morphology and Hypothesis Test

Video object segmentation is the one of the focus in the field of image and video processing.This paper first use s the morphological operators and improved watershed algorithm to segment the image sequences in the spatial do-main,then does the inter-frame change detection using F-detection method and get the initial change detection mask(CDM)through combining temporal and spatial segmentation.With the post-processing method based on the binary mor-phological operators,the final segmentation result can be gotten.The whole process is the morphology processing based on the gray-level image and binary mask,and is simple and easy to implement.The experimental result shows the ef-fectiveness of the method that it can well separate the foreground and background,preferably locate and segment the video object.