Similarity Retrieval of Trademark Images by Vector Graphics Based on Shape Characteristics of Components

Abstract

This paper presents a similarity retrieval of trademark images represented by vector graphics. In the real inspection for new trademark images, similarity for essential objects which show products or services is evaluated to the registry. Although a similarity retrieval system which evaluates similarity between the essential objects has been already reported with trademark images represented by vector graphics, the existing system does not work well in some cases. The proposed method is an improvement version of the existing system. To enhance performance of the system, in the proposed method, the extraction of feature vectors for an object's contour shape is improved and the method of centroid distance is introduced into the feature extraction besides. Experimental results have shown that the proposed method could evaluate similarity of each essential object in trademarks better than the existing system.