A local descriptor for high-speed and high-performance pictogram matching

Pictogram, a simple picture-based symbol, is widely used to indicate important facilities (such as 'rest room') or important rules (such as 'no smoking'). If a practicable pictogram matching technology is available, it must be useful for many applications. Unfortunately, as far as the authors know, there are no practicable methods which satisfy both high-speed (real-time) processing and high performance (high matching ratio). In this paper, we present a novel local descriptor for high-speed and high-performance pictogram matching under a variety of photographing conditions. The proposed method consists of three modules: projection-invariant values called CRN (Cross Ratio Number), shape description scheme for pictograms, and matching acceleration method based on feature vector classification by using the relationship between contour and convex hull. The experiments show our method outperforms the state-of-the-art methods in terms of matching ratio and computation cost.