Design And Implementation Of Handheld Device Self-Navigation System For Large Image Based On Image Retrieval

As an important part of Chinese cultural construction, there is a large quantity of the two-dimensional plane (large image) exhibits in museum, which contain a wealth of cultural information. How to do the personalized self-navigation for these rich cultural information are one of the hot and difficult problems in the field of the current museum digitized exhibition areas. The design and implementation of handheld devices self-navigation system for large image based on image retrieval, and the research of the key technology are introduced in this paper.The method of image similarity measure based on the discrimination and spatial information is introduced. For deleting the feature points with low discrimination, the discrimination of every point is quantified through the calculation of the Euclidean distance of the global feature points. Establish the two-dimensional plane coordinate system of the large image, using the idea of statistics sort to do the image similarity measure through determine the hot zone location information offline and RANSAC algorithm.The method of the image features hierarchical index for a single large resolution image is introduced. This method using the SIFT operator establishes the image hierarchical indexing structure using the global position information of the image. Combined with the LSH and location-based KD-Tree, the range of the search of the KD-Tree is narrowed with the help of the rough location information returned from the upper LSH. This method could help to improve the speed and accuracy rate of the retrieval.The handheld devices self-navigation system for large image based on image retriecal is designed and developed. The system uses the server-client model, and the handheld devices are used for the client of image acquisition and information displaying. The server includes the HTTP based network server and the RTSP based streaming media server, which could provide the services of partial image retrieval and multimedia information displaying for multi-client.