Research Of Image Retrieval Methods Based On Content

Content-based image retrieval is kind of image retrieval technology which utilizes version characteristic(color,texture,shape). CBIR grows as study hotspot along with the progress of computer science and database technology.The article firstly reviews the theory of content-based image retrieval and the development progress, summarize the content-based image retrieval (particularly based on the contour), and discuss about some key technology of which.The article makes the academic researches on the acquisition and retrieval of low-level contour characteristic, based on which a pithiness,effective,practical image retrieval system is developed. The article includes:1 .A new contour segmentation algorithm is discussed, in which the contour of the image is precisely segmented by the inflexion, the shape characteristic is extracted by edge tracing, and the contour of the image is identified by constructing the concavo-convex degree and average field angle; thereby the contour is effectively represented.2.The semantic information and the contour characteristic of the image is considered as synthetical index based on the current shaped-based image retrieval; the subjective semantic and low-level characteristic are inosculated, so as to improve the correlation degree between the low-level characteristic and high-level sematics, and to construct the retrieval frame. Construct the image index to retrieve the image based on the figure contour; the experiment shows that the kind of retrieval system do well in the retrieval of complex image.3.The new measurement of comparability of the image is given, in which different measurement function of comparability is constructed to match the comparability of the image.The different formats of image is retrieved by the above algorithm in the end of the article, and the result shows that the algorithm improve the efficiency and accuracy of the retrieval of complex image and has the steady performance.