Applications of Bilateral Filtering for Image Denoising and Enhancement
Shenghua Teng, Taoi Hsu · 2009
Image denoising and contrast enhancement are important techniques in image processing. Many imaging applications require the image smoothing or denoising while preserving the detailed edges. Simple smoothing operations such as low-pass Gaussian filtering which does not consider intensity variations tend to create blurred images. Bilateral filtering (BF) extends the Gaussian smoothing technique by using a locally adaptive recovery algorithm. The filter coefficients are weighted with corresponding image pixel intensities. It is essentially a convolution process using a nonlinear Gaussian filter, with weights based on pixel intensities. The non-iterative bilateral filter can detect and preserve image edges and average only those pixels on the same side of an edge. It has been reported to have a better edge-preserving capability than linear filters in certain applications. This paper presents a review of the algorithm, characteristics, and imaging applications of the bilateral filter. Experimental results and evaluation study on the performance of bilateral filtering are also presented.