基于块匹配和三维滤波的相干激光雷达图像去噪研究  

Study on Coherent Ladar Image Denoising by Block-Matching and 3D Filtering

作　　者：夏志伟[1] 李琦[1] 刘正君[1] 王骐[1] 

机构地区：[1]哈尔滨工业大学可调谐激光技术国家级重点实验室,黑龙江哈尔滨150081

出　　处：《中国激光》2012年第B06期434-440,共7页Chinese Journal of Lasers

摘　　要：针对相干激光雷达距离像和强度像的不同噪声特性，分别采用基于块匹配和三维滤波（BM3D）算法对距离像和强度像进行了去噪处理，并与非局部均值（NLM）算法的去噪结果进行了比较和分析。实验仿真结果表明，对于强度像，同态BM3D算法明显好于同态NLM算法，在保持目标区域的灰度分布均匀性、保持及恢复图像边缘等方面具有较好的性能；对于距离像，由于一般更注重保持正确的距离值，BM3D和NLM算法均不够理想。Due to the different characteristics between the range and intensity images of ladar, block-matching and 3D filtering （BM3D） is used in both range image and intensity image denoising respectively. And the denoising results are compared and analyzed with those of non-local means （NLM） filtering. The results of experiments show as follows. For intensity image, homostasis BM3D algorithm, which has a better performance in restoring the edge of objects and keeping smoothness in homogeneous area, is better than homostasis NLM algorithm. Generally speaking, preserving right range value is significant for range image processing, so both BM3D and NLM are not sufficient ideal under the circumstance.

关 键 词：图像处理 图像去噪 块匹配和三维滤波 非局部均值算法 相干激光雷达 

分 类 号：TN958.98[电子电信—信号与信息处理] TN911.73[电子电信—信息与通信工程]