Share Email Print
cover

Proceedings Paper

A model-based denoising adaptive diffusion method based on multi-scale bilateral filter
Author(s): Yu Hui; Jun-Sheng Wu; Bin Yu; Chen Zhang
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

In order to filter out the noise points of the medical spine model effectively and keep the details of the model better, a model-based denoising adaptive diffusion method based on multi-scale bilateral filter is proposed. This method firstly extracts the contour lines of the 3D model of the spine by using bilateral filter in multi-scale conditions, and then designs and improves the adaptive diffusion coefficient so as to optimize and control the whole diffusion process. Then, according to the discreteness of the image, the corresponding discrete iterative equations are established to discretize the iterative process, and the iteration stopping criterion is designed to make the denoising and smoothing image model stop the iteration when the correlation between that image model and the noise is the minimum, and finally to establish the spine image denoising model. After compared with the experimental results of the classical PM method, Catte method and other methods, this method achieves good filtering effect in denoising, and also preserves the edge detail features of medical image, much better than traditional filtering algorithm.

Paper Details

Date Published: 8 March 2017
PDF: 13 pages
Proc. SPIE 10255, Selected Papers of the Chinese Society for Optical Engineering Conferences held October and November 2016, 102552R (8 March 2017); doi: 10.1117/12.2264433
Show Author Affiliations
Yu Hui, Northwestern Polytechnical Univ. (China)
Jun-Sheng Wu, Northwestern Polytechnical Univ. (China)
Bin Yu, Xidian Univ. (China)
Chen Zhang, Xidian Univ. (China)


Published in SPIE Proceedings Vol. 10255:
Selected Papers of the Chinese Society for Optical Engineering Conferences held October and November 2016
Yueguang Lv; Jialing Le; Hesheng Chen; Jianyu Wang; Jianda Shao, Editor(s)

© SPIE. Terms of Use
Back to Top