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.
出处
《中国激光》
EI
CAS
CSCD
北大核心
2012年第B06期434-440,共7页
Chinese Journal of Lasers
