Publications

Abstract : In this paper, a new proposal is reported for image quality enhancement. Here, statistically clipped, bi-histogram equalization-based adaptive gamma correction along with its cosine-transformed energy redistribution is introduced for improvement of low-contrast dark images. This approach computes the clipping limit adaptively by observing stretched histogram bins, mean, and median values for each sub-histogram. Considering the clipping limit as the lowest of these three values, this limit ensures the conservation of information content of the image to a great extent. This adaptive clipping limit selection also resolves the issue of over-emphasization of high-frequency bins during sub-histogram equalization. For harvesting more information and better illumination, gamma correction is imparted by using the adaptive gamma value-set. The corresponding gamma value-set is itself derived from the previously sub-equalized interim intensity channel. In addition to this, two-dimensional (2D) discrete cosine transformation (DCT) for gamma-corrected channel is also employed for incorporating the energy-based textural enhancement framework. Validation is performed here by analyzing the enhancement for various remotely sensed dark satellite images by evaluating standard performance indices.