The Comparison of Image Enhancement Methods for A Realtime Under Water Vision System of The Break Water Implementation

Abstract

The implementation of computer vision in the field of underwater image is currently very popular. However, Indonesia's vision system for breakwater work has not yet been implemented. The work of the vision system is still manual using divers [1]. In addition, the application of computer vision in underwater images has a big obstacle because the resulting image in an underwater environment has poor color and contrast. This poor color and contrast are due to a density nearly 1000 times denser than air. This density directly affects the light transmission, which affects the resulting image, which has less color and contrast than the image taken through the air. So it takes image processing such as Image Enhancement and Image Color Restoration [2]. Another issue is building a real-time system. If the system that is built is not real-time, this makes the possibility of data reading errors being significant, and the movement of information is slow [3]. This research presents methods regarding image enhancement and color restoration built in the Graphical User Interface (GUI). The methods applied in this research are Histogram Equalization (HE), Contrast Limited Adaptive Histogram Equalization (CLAHE), and Multi-Scale Retinex with Color Restoration (MSRCR). Those methods are applied to the images produced by the two cameras and run on the single board processor and laptop for the comparison. The result by using 71 underwater images shows that for Underwater Color Image Quality Evaluation (UCIQE) value of the Histogram Equalization (HE) Method yields 20%, meanwhile for Contrast Limited Adaptive Histogram Equalization (CLAHE) yields 8%, and the Multi-Scale Retinex with Color Restoration (MSRCR) method yields 72%.