VIDEO OBJECT SEGMENTATION FOR CONTENT-BASED VIDEO CODING
一种面向对象的视频分割算法

This paper deals with automatic object segmentation method for object-oriented video encoding. Since methods of mathematical morphology are capable of detecting geometrical features such as size, contrast, or connectivity which are important to image segmentation. Intra-frame segmentation is in first step performed using morphological tools. After obtaining different areas in intra-frame segmentation step, a few video object areas can be merged according to their motion parameters. Finally, a new technique for the detection of motion of a local area is proposed. A two-dimensional binary model can be obtained from the object of interest. By introducing a refinement step, the border pixels can be categorized into one of the neighboring objects so as to achieve a minimum predictive error that computed from its model parameters. Experimental results demonstrate a successful segmentation of objects located in a still or moving background.