摘要
基于粒子滤波在非线性非高斯情况下具有较好的预测结果,本文提出了一种自适应背景图像分割新算法,该算法利用粒子滤波对下一帧的前景区域进行预测,进而计算出下一帧各像素点属于背景的概率以指导下一帧图像分割;在前景像素值与背景像素值相近的情况下利用先验知识进行图像分割是一种较好的方法,本文以粒子滤波预测结果与先验概率模型计算结果的均值作为当前像素点属于背景的概率来进行图像分割,实验结果表明,该方法在背景变化范围较大的情况下,可以减少前景点误分割为背景点的概率.
A new adaptive algorithm is proposed by taken advantage of SMC （Sequential Monte Carlo） which have better predictive results under the condition of nonlinear non-Gaussian. The algorithm uses particle filtering to predict an anticipated fore-ground district for a coming flame. Moreover, it calculates the probability of pixels to be part of background in the coming frame to guide image segmentation. It is a good method to segment image on the setting where the pixel values of foreground similar to the ones of the background by using prior knowledge. This paper uses the probability of pixels to be part of background which is calculated by the average of the predict results of particle filtering and the calculated results of prior probability model to segment image. Experimental results show that the proposed algorithm can reduce the error of the pixels of foreground to be segmented as pixels of background compared with 3σ rule when changes in background occur quickly.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2007年第8期1533-1537,1547,共6页
Acta Electronica Sinica
基金
国家自然科学基金(No.60573079No.60673093)
长江学者和创新团队发展计划
怀化学院计算机应用技术重点学科项目
关键词
粒子滤波
先验概率
自适应分割
运动检测
高斯模型
SMC（Sequential Monte Carlo）
prior probability
adaptive segmentation
movement detection
Gaussian model
