摘要
文中提出了一种新的阈值化方法用来在自适应背景的应用中把运动物体从景物中分割出来。传统的方法是用一个简单的阈值来分割物体,但是其中存在一个问题就是难以取得一个恰当的阈值在误警率和探测率间取得平衡。文中这种新的方法的提出在减少误警率的同时避免了失检,此方法基于阈值滞后的概念,采用多次子采样在不同的应用层中获取不同的阈值,在不同的应用层中利用不同的分布概率下的阈值来减少误警率和失检率。实验证明该方法较其他方法分割运动物体更为有效。
This paper presents an new method of thresholding for moving object segmentation in adaptive background update application. In traditional ways, a simple threshold is applied to segmentation the moving object fiom the scence, but there is a problem that a proper threshold value faces a trade-off between false alarms and misdetection. In this paper, a new method of thresholding for moving object segmentation is proposed to avoid misdetection while reducing false alarms. This new approach is base on the thresholding with the hysteresis, and adopts the multi-subsampling to achieve difference threshold value in the application. In these application, we utilize the threshold values to reduce false alarms and misdetectoion. The experiment shows that the approach is more effective compared with the other algorithm to segmentation the moving object.
出处
《微计算机信息》
2009年第6期301-302,共2页
Control & Automation
关键词
多阈值
目标分割
子采样
Multi-threshholding
Object Segmentaion
Subsample
