Trademark Image Retrieval using Color and Shape Features and Similarity Measurement

Worldwide, the number of trademarks is growing exponentially, but it is still a difficult challenge to evaluate perceptual similarity and determine an acceptable measure of similarity for trademarks. To secure the legal rights of a registered trademark, a reliable and an efficient system for the retrieval of trademark images is necessary. We suggest a new method to retrieve trademark images by integrating color and shape features and for similarity measure we use Euclidian distance. Color feature gives the trademark which is similar in color space for a given query trademark and shape feature gives the trademark which is a transformed form of query trademark. Proposed method is tested on 500 trademark images which are downloaded from internet. Our results show that the combination of features gives the effective and efficient results with high precision rate.