Interactive mobile information services

Visual Search technology helps people search for the information through the input of an image rather than words. Image processing and pattern recognition techniques are often involved in recent visual search engines. With the rapid progression of mobile technology, it becomes possible to port the visual search system onto mobile devices. In this final year project, an interactive mobile visual search system was designed and developed on the Android platform, after a brief study of the state-of-the-art algorithms on image recognition. Server–client architecture was adopted as the main framework of the whole system. The system is able to recognize the image uploaded from the mobile device using scalable recognition algorithms with Bag of Words (Bow) model, Scale-invariant feature transform (SIFT) descriptor and Vocabulary Tree. The information relevant to the query image will be transmitted back to the client device after the analysis is done on the server. Location-based social interactive services are also embedded in the client application. Several development methodologies were involved during the system design process in order to find the best approach to maximize the system performance. The latest version of the system implements the algorithms in C programming language with the image database stored on the server. It can run smoothly on the latest Android mobile phone in a local area network environment. Reasonable processing speed and accuracy were achieved as well. A short video showing several demos of the system can be found at: http://www.youtube.com/watch?v=wdwn-w9cZOMBachelor of Engineerin