摘要
提出了基于递归高阶统计的运动目标分割方案 ,主要用于会议电视图像序列 .与传统的高阶统计方法相比 ,基于递归高阶统计的分割方法从高斯噪声背景中提取信息 ,由于利用了前 n帧图像的信息冗余度 ,递归高阶统计方法可以更有效地减少噪声、检测小目标 ,并且提高了目标的分割准确度 .模拟试验结果显示 。
This paper proposed a moving object segmentation method based on recursive high order statistics(HOS) used in videoconference sequence. The recursive high order statistics method can extract the signal from Gaussian noise. Because of using the former n image information, it can obtain such better estimates of HOS as the reduction of additive noise, detection of the small object and well performance of the segmentation compared with the classical higher order statistics method. The simulation test indicates the segmentation results are good.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2000年第7期991-993,共3页
Journal of Shanghai Jiaotong University
关键词
图像分割
会议电视图像序列
运动目标分割
image segmentation
image compression
recursive higher-order statistics
morphology filter-ing
