4 January 2013 Ladar range image denoising by a nonlocal probability statistics algorithm
Author Affiliations +
Abstract
According to the characteristic of range images of coherent ladar and the basis of nonlocal means (NLM), a nonlocal probability statistics (NLPS) algorithm is proposed in this paper. The difference is that NLM performs denoising using the mean of the conditional probability distribution function (PDF) while NLPS using the maximum of the marginal PDF. In the algorithm, similar blocks are found out by the operation of block matching and form a group. Pixels in the group are analyzed by probability statistics and the gray value with maximum probability is used as the estimated value of the current pixel. The simulated range images of coherent ladar with different carrier-to-noise ratio and real range image of coherent ladar with 8 gray-scales are denoised by this algorithm, and the results are compared with those of median filter, multitemplate order mean filter, NLM, median nonlocal mean filter and its incorporation of anatomical side information, and unsupervised information-theoretic adaptive filter. The range abnormality noise and Gaussian noise in range image of coherent ladar are effectively suppressed by NLPS.
© 2013 Society of Photo-Optical Instrumentation Engineers (SPIE)
Zhi-Wei Xia, Zhi-Wei Xia, Qi Li, Qi Li, Zhi-Peng Xiong, Zhi-Peng Xiong, Qi Wang, Qi Wang, "Ladar range image denoising by a nonlocal probability statistics algorithm," Optical Engineering 52(1), 017003 (4 January 2013). https://doi.org/10.1117/1.OE.52.1.017003 . Submission:
JOURNAL ARTICLE
12 PAGES


SHARE
RELATED CONTENT

A new algorithm for image denoising
Proceedings of SPIE (November 24 2009)

Back to Top