Contrast Enhancement Using Novel White Balancing Parameter Optimization for Perceptually Invisible Images

Abstract

A novel white balancing algorithm is proposed in this paper to automatically enhance the global contrast degraded imperceptible images. The technique is applied on four publicly available image dataset, CSIQ, KADID, TID and SIPI. Colour images consist of three channels viz. Red, Blue and Green. A contrast degraded colour image visually appears similar to an image with one or more distorted channel. 12 images are obtained by enhancing one channel of the contrast degraded image at the cost of other channel using White Balancing algorithm. Four images with best quantitative performance metrics, visual similarity index (VSI), gradient magnitude similarity index (GMSD), patch-based contrast quality index (PCQI) and peak signal-to-noise ratio (PSNR) determines the pair of weak and prominent channels. An optimization algorithm then enhances these channels and the image with the best quantitative performance metrics is chosen as the enhanced image. Quantitative and qualitative results demonstrate that the proposed method produces an enhanced image with superior perceptual quality, and gives the best average results for all the parameters across every dataset as compared to the state-of-the-art methods.