Content-based color trademark retrieval system using hit statistic

Abstract

This paper proposes a content-based color trademark retrieval system. First, the edges of each color trademark are detected. Then, the smallest circle that covers the trademark is derived. Based on the smallest circle and edges, the feature called hit statistic on a dartboard of the color trademark is extracted. Using this feature, this paper constructs an efficient and simple color trademark retrieval system, which is robust to rotation, translation, scaling and some geometric distortions. Some experiments are conducted to compare the proposed system with the existing one using Zernike moment, and the results show that the proposed system is superior to that using Zernike moment.