Font Size: a A A

Research And Implementation Of Building Recognition Algorithm Used In Augmented Reality Systems

Posted on:2014-01-01Degree:MasterType:Thesis
Country:ChinaCandidate:Q F QiFull Text:PDF
GTID:2248330398972228Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
With the rapid development of multimedia information processing technology, more and more multimedia techniques have emerged, making the access to achieving daily information increasingly visualization and efficient. The augmented technology could combine the improved information with the real pictures and satisfy the requirement of getting the vivid information easily. Content-based image retrieval technology can achieve the image similarity judgments, widely used in image management, image retrieval and other aspects.this paper applies the Content-based image retrieval technology in the architecture image retrieval innovatively with the building identification C/S system, which enable people using mobile device to get the real information of the building and complete the information processing module of the system.This paper firstly analysis the method of Content-based image retrieval technique in feature extraction and retrieval and compare the accuracy and matching speed differences between the various characterization methods. Test the methods such as color histogram, color moment, domain color, texture based on the binary pattern, SIFT and SURF. In these methods, color histogram, SIFT and SURF can meet the accuracy requirement, but with SIFT, the search time is too long, and in SURF and color histogram it is ok.We select several features according to the accurate matching and good real-time principle in order to combine the building identification algorithm and put it into practice with the buildings of BUPT. The process of the building identification algorithm is, firstly, do image segmentation preprocessing, then use the color histogram narrow the search range, finally, use the SURF algorithm for exact result. The building identification algorithm is the series connection of image segmentation, color histogram and the SURF algorithm. The test results show that the building identification algorithms can achieve accurate and effective building image retrieval in time.This paper complete the building identification C/S system,achieve the access of mobile device to PC server and display the PC feedback information.This building identification system use the mobile device collect the building image and send it to PC server.the PC server complete the retrieval with the algorithms and give the mobile device feedback information through the whole operation process of the test system takes about15s.
Keywords/Search Tags:Augmented Reality, Content-based image retrievaltechnology, Image feature extraction and matching, buildingidentification algorithm
PDF Full Text Request
Related items