摘要
提出了应用正则粒子的再采样算法来实现红外图像的消噪处理。首先从先验概率密度的加权随机粒子重建它的后验密度函数,依据核带宽从近似的后验密度函数中随机重新采样,得到新的正则粒子,对粒子权值更新,进行归一化处理;然后对正则粒子滤波器进行优化,预测状态粒子以及均值和方差,最后在监测区域中构造物体的运动模型。通过对具体红外成像的处理分析,在增加粒子数目和重采样结合的情况下,红外成像识别效果比较清晰,这样体现了正则粒子用于红外成像消噪处理的有效性。
Using the particle resampling algorithm to achieve infrared image denoising processing is put forward.First,from a prior probability density of the weighted random particle,its posterior density function is rebuilt.Based on the nuclear bandwidth from the approximate posterior density function of random resampling,a new regularized particle can be obtained.The particle weight value can be updated,and the normalized treatment was performed.Then the regularized particle filter was to be optimized,and the state particles and the mean value and variance can be predicted.Finally the movement model of tectonic objects can be mornitored in the end zone.Through the processing of specified infrared imaging analysis to increase the number of particles and resampling combination of circumstances,it is identified that the effects of infrared imaging are relatively clear.This reflects the effectiveness for regularized particles used in infrared imaging denoising.
出处
《实验技术与管理》
CAS
北大核心
2010年第4期42-44,共3页
Experimental Technology and Management
关键词
正则粒子
重采样
核函数
regularized particle
resampling
kernel function
