An Improved Offline Stable Point Filtering Method for Mobile Search Application

Mobile visual search business has been prospering since research break-through on content based image retrieval (CBIR) and continuous improving mobile technology. In such search application, user takes photo and uploads photo to server. Retrieval is done at the server side and user gets result from the server. Currently search applications have difficulties. 1) Mobile photos have relatively low qualities, and photos contain various affine transforms. The two drawbacks will influence the retrieval accuracy. 2) Retrieval speed and accuracy could not meet a balance for large scale database. Current index mechanism for high dimensional local feature such as SIFT is not fast enough while lots of low dimensional feature cannot ensure high accuracy. This paper proposes a fully-automated offline stable point filtering method for mobile visual search application. We use various transforms to simulate effects caused by mobile photo images. They are processed in our offline method to reduce the index size of the retrieval application. Experiments show that with proper offline processing, 13.83% memory or disk space could be saved when retrieval application loads SIFT features into memory for LSH at the online stage while application still maintains a high query accuracy.