Study On Object Detection And Location Of Underwater Images

Nowadays, both image technology and image processing system have comprehensive applications in the areas like ocean engineering, aviation industry, biomedicine, communication engineering, military security and agriculture. Image processing technologies take great importance especially upon ocean surveillance, seabed prospect and underwater target detection. As one of the key points in modern ocean high technology, the underwater object detection and location techniques has developed sharply, and it has been applied in many aspects.Under this application underground, we do some research in this technique.This paper is mainly about the research of the underwater object detection and location techniques, it contain three parts: first, edge detection of the underwater objects, second, the extraction of the object center, the third, the location of the object.We first summarize the development of edge detection algorithms, give introduction of some algorithms in details, and take into consideration of the complexity and feasibility of the algorithms. Under the research of the local complexity of image, we propose the adaptive and Multiresolution edge detection algorithm base on the canny operator. The algorithm is used to cope with the two problems existing in the Canny operator, the first is the combination of denoise and location, the second is the adaptive choice of the two thresholds which use to link edge. And the complexities of our algorithm only increase a little. In the second part, we propose different algorithm towards different objects in geometric characteristics, especially in the extraction of the center of the arc object, we propose the overlapped segment edge fitting algorithm, this algorithm divide an arc into several small segments, and get the parameters of each segment, at last we use the clustering method to get the result. In the end ,we do some research on the object location technology, the location algorithm is concern with the camera calibration,so we give introduction about some calibration algorithm,we modify the calibration algorithm based on plane,after that we use the projective matrix to locate the object, and also we give error analysis.