Underwater Image Enhancement Using Neighbourhood Based Two Level Contrast Stretching and Modified Artificial Bee Colony
Sourav De, Sandip Dey, Shouvik Paul · 2020 IEEE 7th Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON) · 2020
Underwater images usually endure from colour deformation and low contrast, since light is dispersed and absorbed when it passes through water. In this paper, a new two-level, neighbourhood based contrast adjustment method is proposed for underwater image enhancement, which can be effectively applied to put back and enhance the quality of underwater images. Several factors typically influence to reduce the lighting in water, that may lead to poor visibility of objects in underwater images. In this method, initially, the pixels are segregated into two halves from the image histogram. A Modified version of Artificial Bee Colony (ABC) algorithm is proposed to identify points in the histogram, based on which, the pixels are stretched in both sides to increase its construct level. The supremacy of the proposed method is visually and quantitatively established in reference to the optimum fitness, mean fitness, Structural Similarity Index Measure (SSIM) and Underwater Colour Image Quality Evaluation (UCIQE), average SSIM and average UCIQE values for all test instances. Finally, Friedman test and two tailed t-test have also been conducted to statistically establish the efficiency of the proposed method. Experimental results demonstrate that the proposed method outperforms others.