Image Matching Technology Based On Multy Underwater Camera

With the developing of the detection about marine resources,more traditional energy and new energy was found under the ocean, moreand more countries have improved the strategic position of the ocean to ahigher level. In order to better detect oceanic resources and developMarine resources, people have to deepen the understanding about theMarine. ROV with its unique location and easy handling gets the favor ofresearchers.In view of the underwater robot, this article puts forwards a kind ofimage matching system which is available and stable. This system will bemore efficient and convenient for underwater image matching and rebuildthe underwater image.This article introduced the types and development of the ROV, and thenexpounded the necessity to do the research in this article. Then design amatching system on CCD after illustrating the preprocessing, featurepoints matching, image matching and analysis.In the preprocessing stage, the median filter and mean filter arecontrasted. Remove the salt and pepper noise. On the selection of featurepoints, this article analyses the Harris algorithm and eventually selects theSIFT algorithm. Due to the uniqueness of the feature descriptor, this algorithm can match the images more stable and reduce thecomputational cost. The SIFT algorithm realizes the rotating, zoominginvariant, and can adapt to a certain feature extraction of illumination andthe change of viewpoint.The RANSAC and the nearest neighbor analysis are used to reduce theerrors in the matching, which improved the accuracy of the match. Aftergetting the feature points this article adopts the way of the matrixtransformation to joining the images together using the Matlab.The experiments show that the system has good robustness andautomatic matching capability.