Adaptive geometric features based filtering impulse noise in colour images

An adaptive geometric features based filtering (AGFF) technique with a low computational complexity is proposed for removal of impulse noise in corrupted color images. The effective and efficient detection is based on geometric characteristics and features of the corrupted pixel and/or the pixel region. A progressive restoration mechanism is devised using multi-pass non-linear operations. Through extensive experiments conducted using a wide range of test color images, the proposed filtering technique has demonstrated superior performance to that of well-known benchmark techniques, in terms of objective measurements, the visual image quality and the computational complexity.