摘要
提出一种基于时空分割融合的视频对象提取改进算法:时间分割基于变化检测,其关键是通过直方图分析选取合适的阈值;空间分割采用修正的分水岭变换,并通过一种基于时间距离和空间距离测度的区域合并算法克服“过分割”问题。实验结果表明,它可取得比COST211AM更好的分割结果。
In this paper,a video segmentation algorithm based on fusion of spatio-temporal segmentation is proposed.Tem-poral segmentation is based on Change detection.Its key is the selection of appropriate threshold,which is obtained by his-togram analysis.Improved watershed is utilized for spatial segmentation,and a region-merging algorithm based on novel tem-poral distance and spatial distance is adopted to overcome ″over-segmentation″.Experimental results show that the proposed approach can achieve more desirable results than COST211AM.
出处
《电视技术》
北大核心
2003年第9期26-28,39,共4页
Video Engineering
