Contrast Enhancement of Dark Images using Stochastic Resonance in Wavelet Domain

Abstract

—In this paper, a dynamic stochastic resonance (DSR)-based technique in discrete wavelet transform (DWT) domain is presented for the enhancement of very dark grayscale and colored images. Generally in DSR, the performance of an input signal can be improved by addition of external noise. However in this paper, the intrinsic noise of an image has been utilized for the purpose of contrast enhancement. The DSR procedure iteratively tunes the DWT coefficients using bistable system parameters. The DSR-based technique significantly enhances the image without introducing any blocking, ringing or spot artifacts. The algorithm has been optimized and made adaptive. Performance of the given technique has been measured in terms of distribution separation measure (DSM), target-to-background enhancement measure based on standard deviation (TBEs) and target-to-background enhancement measure based on entropy (TBEe). When compared with the existing enhancement techniques such as histogram equalization, gamma correction, single-scale retinex, multi-scale retinex, modified high-pass filtering and Fourier-based DSR, the DWT-based DSR technique gives better performance in terms of visual information, color preservation and computational complexity of the enhancement process.