An indexing scheme for efficient camera angle invariant image retrieval

In specialized applications such as security, monitoring and surveillance, camera angle invariant image retrieval might be very useful. Typically, for shape-based angle invariant image retrieval, we need an object image at every angle where the shape of the image changes. This will generate a huge size of image database. Even worse, if we simply apply existing indexing schemes to this database, we will face serious performance problems due to the data size and data redundancy. In this paper, we propose a new indexing and matching scheme for shape-based angle invariant image retrieval, focusing on two types of camera movements: horizontal rotation and fixed rotation. To reduce the space requirement for indexing the huge database, we use two levels of indexing structures: (i) static indexing for horizontally rotated images for horizontal angle invariance, and (ii) dynamic indexing for fixedly rotated images for rotation invariance. Through the experiment, we show that our proposed scheme can achieve much better performance than other competing methods.