Wavelet based perspective on variational enhancement technique for underwater imagery

In deep sea environment, the quality of underwater imagery is primarily affected with low contrast, blur and color cast due to the absorption and scattering. In order to deal with these discrepancies a framework is proposed in this paper wherein a set of energy functionals is applied on the approximation and the detailed coefficients of the image. The approximation coefficients of RGB components are modified for adjusting the average intensity value of the image followed by the color correction of these coefficients at finer scales. Subsequently, the processing of detailed coefficients are done for improving local contrast of the image. The performance of the proposed method is evaluated qualitatively and quantitatively on three underwater datasets at varying depths. Qualitative analysis is carried out by comparing the hue histogram of input and output images, whereas quantitative analysis comprises of PSNR, Entropy and SSIM quality metrics. The results of the proposed method are compared with state-of-the-art methods. From the obtained outcomes, it is observed that the proposed method significantly removes the color cast by improving the contrast of underwater images in addition to preserving its detailed structural features.