MOSIR: Mobile-Based Segment and Image Retrieval

This paper proposes a content-based image retrieval system for mobile phones, named MOSIR. The system facilitates instantaneous search on mobile phones for images similar to photos taken via phone cameras. Edge-based and color-layout features are used to find similar images. The system also enables region-based queries by detecting salient regions and extracting their features. To overcome security problems, the system applies both self-defined email and web application software. The system overcomes limitation of display panels and control key in common mobile phones by using Flash-Lite as a new user interface approach to view images of different sizes and contexts. The system was tested with 2,250 images, and results were satisfying. Therefore, MOSIR is a promising system to perform instant image search on mobile phones. Currently, it is targeted for common mobile phones. Eventually, MOSIR is aimed to support smart phones and retrieval of semantically related information to resulted images.