Research On Underwater Image With Complex Background Enhancement

Machine vision systems are widely used in modern industrial production, military reconnaissance and other scientific fields. However, in the images which collected in the water background, due to the special optical properties of the aqueous medium and the complexity of the underwater environment to detect and identify the visual system has brought great difficulties. To solve this problem, this paper has been study on the water background images enhancement; especially for underwater video image enhancement method of how to remove movement background influence research.This paper analyzes the optical properties of aqueous media and imaging characteristics, and then introduced the commonly used classic tools in enhancing underwater image quality and processing of the application process results. Secondly, using of all kinds of image evaluation function was evaluated, the result of the corresponding methods of treatment evaluation results prove the validity of the proposed enhancement method.In bubble movement information of underwater video image, the traditional means of enhance cannot achieve a satisfactory requirement, thus improved independent component analysis algorithm is proposed in this paper to deal with it. The algorithm separates the background and moving targets in image sequences, and then separates bubbles and other moving targets to remove air bubbles on the gray domain. Finally, in the image which has been wiped off the air bubbles assign its adjacent sequence of video image in the same position but clear pixel values. And the background of bubble edge information extraction and separation components made in the process of optimization, compared with the commonly used method, has reached the requirement of accuracy and real-time performance.At the end of the paper using optical flow method can accurately detect the features of the sports ground of processed image has carried on the analysis and calculation, the result proves that the reliability of the method proposed in this paper.