Nonlinear diffusion filtering method for flash X-ray CCD images
Jinming Chen · Guangdian gongcheng · 2006
According to the characteristics of flash X-ray radiographic CCD images, a new diffusion filter called median-nonlinear diffusion filter (MNDF) was proposed in the paper. This method estimates image edges based on median filtering image. Then, by solving a PDE (partial differential equation), a filtered image can be derived. From which, the nonlinear advantage of the median filters and the nonlinear diffusion filters can be utilized to eliminate noise and preserve edges more efficiently. Experiments show that the method can eliminate the combined noises of Gaussian noise and the impulse noise efficiently while preserving edge details. The PSNR (Promotion of Signal-to-Noise Ratio) of the new method is 3~5 times higher than that of traditional methods and its MSE (mean square error) is 1.3~2.7 times smaller than that of traditional method. Good denoising results for flash X-ray radiographic CCD images are obtained with image edges preserved.