Research On The Key Technologies Of Detection Underwater

The thesis comes from the national technology based plan topic-The Technology of Approaching Targets in Emergency on Maritime Fast. In order to locate the submarine target in danger precisely, the thesis mainly do two parts:one is to recognize the target from the sonar image, the other is to enhance the optical pictures captured submarine.According to the feature of sonar image token by high-resolution sonar, the thesis decides to recognize the target using Level-set and SVM based algorithm. According to the feature of optical image token underwater, the thesis decides to enhance the images using Dark Channel Prior and Retinex based algorithm.Level-set and SVM based algorithm in this paper extracts contours by using level-set, calculates invariant moment of the contours as the feature vectors and then inputs the feature vectors to the support vector machine(SVM) that have been trained to obtain the recognition results. This algorithm combines the merit of the level-set, the invariability of shifting, rotation and scaling and the unique advantage of the support vector machine (SVM) in small samples and nonlinear pattern recognition. The results of experiments show that the proposed method has higher recognition rate and lower false-positive rate.Due to the absorption effects, scattering effects and the convolution effects to light underwater, the images shoot underwater has a serious non-uniform brightness and the details of the images is fuzzy. Usually the signal-to-noise ratio of image is low and the contrast becomes very poor. The algorithm of Dark Channel Prior is raised by MingKai He last year, which is based on the discipline that most outdoor picture without mist. The algorithm of Retinex is based on the Constancy of color. The thesis uses Dark Channel Prior and Retinex based algorithm to enhance the quality of images shoot underwater and the result of evaluate using subjectively and objectively are very good.