基于正则粒子重采样算法的红外成像消噪处理

 针对红外成像消噪的粒子滤波退化问题，提出正则粒子重采样算法。该算法从粒子群重采样获得粒子云(xjk，nj)mj=1，解决了粒子多样性消失的问题并克服粒子匮乏的现象；接着又通过添加辅助粒子v，将下一时刻观测值权值大的粒子进行标识，使粒子权值ωik∝p(xk｜yik-1)p(xk｜μik-1)更加稳定；给出了运动物体的红外成像消噪模型。实验仿真表明：正则粒子重采样算法通过添加辅助粒子使红外成像消噪效果好，成像清晰度在95%以上。

Infrared imaging denoising processing based on regular granule resampling algorithm

The regular granule heavy sampling algorithm is proposed for solving the deterioration of the granules in the infrared imaging denoising process. The granule cloud is obtained by the algorithm based on the granule resampling which can eliminate the phenomena of the granule diversity vanishing and granule want, the granules with great weight value observed at the next moment are marked to make the granule weight value more stable by adding some auxiliary granules, and then an infrared imaging denoising model of a moving object is established. The experimental result indicates that the method makes the effect of the infrared imaging denoising much better, and the imaging definition above 95%.