Spatial domain color image enhancement based on local processing

摘要

In this paper, a color image enhancement based on local processing approach is proposed. The proposed algorithm includes image enhancement in spatial domain. Image is treated locally, means enhancement operator is applied only where it is needed. Proposed method takes place in three steps; initially luminance of image is enhanced by a non-linear transfer function. In second step, contrast enhancement is carried out using neighbourhood dependent approach and in the third step; tone reproduction is carried out by 2-D local Gamma correction. In this technique Hue and Saturation is preserved because the method is applied only on V (luminance) part of an HSV (Hue-Saturation-Luminance). Finally, Hue (H) and Saturation (S) component of the original image and enhanced Value component (V) are combined together and converted back to an RGB image. This algorithm is very effective for non-uniform illumination images. Performance of image enhancement is evaluated through visual perception, colorfulness index and number of saturated pixels (percentage). The experimental results validate that images improved by proposed algorithm are clearer, natural looking in comparison with other techniques of image enhancement.