Image Stitching Machines Fish Dual Camera Real-time Target Detection System

With the advancement of science and technology, people’s desire for knowledge has become increasingly stronger and stronger. From the vastness of space to the mysterious ocean, the range of the study on these fields become wider and wider and the level of these study become higher and higher. Especially the mysterious ocean attracts a growing number of countries and academics involved. In the background that "Jiaolong" constantly refresh the diving depth record, making ocean exploration becomes more attractive than before in recent years. But as research becomes more complex and deeper, more intelligent and utility equipments have become more and more important for water research. The real-time panoramic vision system will be widely applied in the safety of equipment and personnel, to be a powerful assistant and to simplify the research projects.Its market value is limitless.Video is made up of a series of frames that associated by the same time intervals. In order to make computer real-time process the frames in line, we need to connected different frames by some methods, and among them image matching fusion method is a good way. What’s more, how to real-time fuse the image frames which is collected by multi cameras, how to effectively accomplish image processing and target detection on these fused images, and so as to obtain the useful information to provide to the system operator, are required parts that need to be fully considered of a real-time vision system. The Aspects mentioned above are hotspots and nodi in vision research field these years.This paper mainly relying on the robotic fish platform with one collect camera which is designed by the Peking University. By applying Image mosaic method and the motion recognition method used in this paper, we upgrade the vision system into to two collect cameras, and make the system successfully real-time process the frames and detect the moving robot fish. Though there are a lot more work to do, like to make it more intelligent, to make the system functioning more effective in order to be better applied when used with tracking methods or deeper data mining application.The last part is to summarize the work we have done, and look forward to the future of the research we devote.