光学图像复原的质量评价与分析  

Quality evaluation and analysis of optical image restoration

作　　者：陈丽娟[1] 汪锋[2] 

机构地区：[1]景德镇学院,江西景德镇333000 [2]中国人民银行景德镇市中心支行,江西景德镇333000

出　　处：《激光杂志》2016年第9期133-137,共5页Laser Journal

基　　金：江西省教育厅科技项目(GJJ09529)

摘　　要：针对光学图像特征的先验规则不足和干扰较大时,对图像复原的全息性不好,复原质量不高的问题,提出基于高阶矩纹理子空间多维谱峰搜索的光学图像全息复原方法,采用连续小波降噪方法进行图像的干扰特征滤波,在光学图像的纹理子空间中进行高阶矩特征提取,结合图像的先验规则,进行多维谱峰搜索,实现图像特征点的自适应匹配标定,实现光学图像的全息复原。仿真结果表明,采用该算法进行光学图像复原的质量较好,峰值信噪比较高,降低了均方根误差,计算开销较小,指标性能优越。Insufficient prior rules for optical image characteristics and interference are bigger, the image restoration of holographic nature is bad, to recover the issue of quality is not high. Proposes a texture subspace based on higher order moment of multidimensional spectrum peak search method of holographic optical image restoration, continuous wavelet noise reduction method is adopted to improve the image characteristics of interference filter, in the optical image texture subspace of higher order moment of feature extraction, combined with the prior rules, of the image of multidimensional spectrum peak search, realize adaptive image feature point matching calibration, realize optical image holographic reconstruction. The simulation results show that the algorithm is adopted to improve the quality of optical image restoration is better, the peak high signal noise ratio （SNR）, reduced the root mean square error, less computing cost, superior performance.

关 键 词：光学图像 复原 小波降噪 峰值信噪比 

分 类 号：TN27[电子电信—物理电子学]