Phase Contrast Enhancement Combined with Dark Channel Prior for Underwater Target Recognition

Abstract

Themajor obstacle to the detection and identification of objects in underwaterimages is often due to their poor quality. Underwater images are known to bestrongly degraded by light absorption and scattering. It is therefore veryimportant to improve the quality of the images to facilitate the objectsdetection. For this purpose, we propose to combine a hardware and softwaretechniques for image quality improvement. The first by using a polarimetricimaging optical system to reduce the scattering effects when acquiring images.The second by apply a Dark Channel Prior (DCP) dehazing method to the obtainedpolarized images. This approach has the advantage to be fast and efficient,because it does not require many software processing to improve the imagequality. In this paper we validate our approach by introducing it into acomplete process of target identification based on phase contrast enhancementand target recognition. Experiments carried out in our laboratory have shownthat this approach greatly improves the target identification.