Noise Cancellation Using Adaptive Trilateral Filter

By exhaustive study of image restoration and noise detection, we find Trilateral filter proposed by Garnett, Huegerich and Chui [13] an extension to the bilateral filter [1] [11] shows remarkable result while denoising digital images corrupted with mixed noise. Further Dong, Raymond H. Chan, and Shufang Xu [17] improved this trilateral filter for images corrupted with impulse noise having more than 60% noise ratio. In this work we propose and justify here an algorithm employing two stage trilateral filter, a combination of above mentioned filters, which has an ability to detect impulse noise ratio in mixed noise and restore the corrupted pixel. Following theoretical derivation based on both authors model (Garnett et al. and Dong et al.), we decouple the problem into two phases. First, we identify the candidates, the pixels that are likely to be corrupted by the impulse noise, and the noise ratio. In second phase, the image is deblurred and denoised simultaneously using the trilateral filter [13].Our approach is to dynamically implement ROAD as well as ROLD. In terms of noise suppression and edge preservation , our restored images shows significant improvement over the number of existing techniques. Some examples illustrate the performance of our method.