The Technique of Spatial Relationships-Based and Shape-Based Multi-Feature Combination of Trademark Image Retrieval

Abstract

There is no doubt that the technique of trademark image retrieval based on multi-feature is increasingly popular. It’s proposed that a combination of the feature of spatial relationships and shape-based is used to retrieve trade mark images in this paper. Firstly, angle relations describe the relationship each spatial object and a special similarity measure is used; Secondly, in order to well describe the shape features of the trademark images, a Euler-number of shape-region-based and compactness and convexity based on shape-contour are proposed; Thirdly, the similarity measure of an inquiring image and database images in the total number of objects is applied with the combination of multi-feature. Good results have been obtained in the following experiments, which prove the multi-feature combination way is better than other single-feature ways.