Image Scaling & Bilateral Filtering

Image Scaling and Bilateral Filtering

Course Project: Psych 221: Applied Vision & Image Systems

Haarith Devarajan, Harold Nyikal

Winter 06

Abstract:

Image up-scaling is a transformation that involves up-sampling and interpolation of images. Interpolation filters by their low-pass nature result in blurry edges in an image. It is desirable to be able to blow up images while preserving the fidelity of sharp edges. Linear high-pass edge preservation algorithms are popular but are limited in that they are very transparent to noise. In up-scaling noisy images, edge preservation is traded for noise suppression, and vice versa. In this project we propose the use of bilateral filters for noise-suppressing and edge-preserving image scaling. Bilateral filters are non-linear filters with influence functions in both spatial axes and pixel intensity, and have been shown to smooth out images while preserving edges. The image scaling algorithm is a cascade of a bilateral edge-preservation filter, a laplacian edge enhancing filter and a scaling kernel with a bilinear interpolation filter.