A Research On Underwater Target Detection Algorithms Wich Based On Glimmer Laser Image Information Fusion

Underwater target detection has been a serious problem of different countries. The discovery is the key to the success of it. Among this, optical detection equipment by carrying on AUV for the detection and identification has a unique advantage. It is an important way of underwater target detection. So the AUV's light vision system is a important part of finish the detection by AUV. .And also to be the basis for the detection and recognition.This paper analyzes the characteristics of underwater images first,describes the mechanism of shimmer and laser imaging cameras for mines and circumstances of the mine image capture experiment in the pool. Meanwhile it also shows the experiment results.Then based on data obtained in experiments on pond, the application of traditional methods of pretreatment and edge detection algorithm to the glimmer / laser image processing have been done. From the result, we can see the common image enhancement method can improve the image quality to some extent, but the traditional edge detection method to be ineffective. So this paper focuses on the edge detection algorithm based on active contour models in mines edge detection. The results show that, In the glimmer mine image, the method seems to be effective. But in the laser image,this method can't get a very good result.In traditional GVF-Snake active contour models,the initial contours need to be set by way of human-computer interaction. So it can't satisfy the autonomy of AUV. This paper proposes an improved edge detection algorithm based on GVF-Snake which combined with Candy operator. This method can set the initial contour automatically by the computer. After experimental verification, this method can get a good initial contour which is automatically set by the computer in the laser image. But in the glimmer image, because of the noise, the initial contours can't be successfully set.Improved GVF-Snake algorithm can achieve ta good results in the glimmer image edge detection image, but can't set the initial contour automatically, meanwhile in the laser image,the algorithm can set the initial contour automatically,but can't get a good edge detection result. To solve this problem,this paper propose a mine edge detection algorithm based on the glimmer / laser image information fusion.This algorithm can satisfy the autonomy requirements of AUV and get a good result of mine image edge detection at the same time. At last, this paper explore the refinement method based on edge contour, and got the binary image which just keep the target contour information and with the contour width is one. The binary image can be used to target recognition.