Smartphone Book Image Recognition System Based On Vocabulary Tree

With the upgrade of smart phone and highly development of mobile communication technology, smartphone has brought people a new experience of surfing Internet nowadays. With the increasing of image multimedia information, traditional text retrieval engine has not well meet the retrieval need of users, so image retrieval engine has been developed quickly these years. Book image retrieval application has been widely used among these applications. The book image retrieval applications based on smartphone were generally retrieval according to the barcode and cover image of books, so they can only search one book each time and were lack of effective integration of book-related information.In order to remedy the shortage of traditional book image retrieval applications, this paper designed and implemented a smartphone book image retrieval system based on image retrieval algorithm of vocabulary tree by considering the scenarios of books on bookshelf. The system uploaded the picture of books on bookshelf captured by smartphone to server for the purpose of image retrieval, and it also provided more detailed information of books for users by the web crawler system.In order to improve the image retrieval accuracy, the system need to segment out the individual spines in the query photo by analyzing the characteristics of the images of arranged books on bookshelf. By edge detection, calculating angular, removing short edges, detecting long straight lines, this paper realized the method of book spine segmentation and verified the efficiency and accuracy of the algorithm by test.Then, this paper realized the image retrieval algorithm based on vocabulary tree. According to the SIFT features extraction algorithms and bags of visual features, by adopting the method of k-means clustering, the extracted image features were generated to visual vocabulary. The method also combined with the method of TF-IDF in text retrieval to improve the efficiency of image retrieval.In order to integrate books information of different web sites, this paper also designed the topic crawler to get the information of books on web. According to the characteristics of information captured and the analysis of web source code, the crawler system realized the function of capturing and storing books information from corresponding websites. By doing this, provided a smartphone book retrieval system combined image retrieval with Web information retrieval.