Isolated noise detection and reduction in color images

A selective median filtering method is developed which successfully detects and reduces isolated noise in an image. In this research, the median filter and gradient edge detection properties and thresholding are combined to develop the selective filtering technique. This filter is compared to two classical median and unweighted average filters, using corrupted images with isolated noise. The selective filter outperforms the classical filters, by demonstrating high noise reduction and high edge preserving properties. The performance measures are based upon mean error calculations at different signal to noise ratios, and psychovisual tests