Research Andapplication In The Feature Point Extraction Of Deep-Sea Creatures Based On The Image Processing Technology

Since the 21 st century, we intensified the efforts on detection of deep sea world. China’s independent research and development of the jiao-long manned vehicle, marks China’s ability to detect deep-sea already can be comparable with the world powers. DSV in detecting deep brings back a lot of film and television information at the same time, the captured data processing has become the subsequent task. After a good deal with image can not only better display of deep-sea exploration results is more important to be able to better the depths of the sea. This paper mainly studies the deep image feature extraction and matching, and application in image matching. Do research work is as follows.First, this paper made a research on the process of image Mosaic, analyzes the of work to do each step and in a component of the whole technology. Through a variety of methods of narrative image splicing technology, coupled with topic, choose the image matching based on feature points.Fish and then, the deep image, image as an example, the pretreatment research, on the optical properties in the water as the breakthrough point, summarizes the characteristics of the deep image. Through the characteristics of the deep sea image illustrates the necessity of image preprocessing, image preprocessing of images and use for which the feature point extraction, is not the ideal extraction result further confirmed that the necessity of image preprocessing. Gray, the histogram of deep sea fish image made change, median filtering, average filtering and gray level transform preprocessing operations.Again the image using Harris algorithm, the improved Harris algorithm and SUSAN algorithm of the three algorithms for fish image feature points extraction respectively. Through the analysis of illustration of the principle and experimental results, comparing the advantages and disadvantages of three kinds of algorithm.It shows that the three algorithms have their advantages and disadvantages, and at the same time, the advantages and disadvantages of the three algorithms are compared.In the extraction of feature points on the basis of deep image feature point matching technology is studied. This article uses the SIFT algorithm and SURF algorithm two kinds of algorithms of feature point matching, this article introduce the principle of the SIFT algorithm, different SURF algorithm and the principle of SIFT algorithm and simulation experiments of sea anemone images, to select the most appropriate deep image matching algorithm.SURF algorithm is more suitable for deep sea image matchingFinally, this article through to the image coordinates transformation to do image stitching and neighborhood average method is used to generate Mosaic image is optimized.