Study And Design Of The Shape-Based Trademark Image Retrieval System

The trademark is popularly named "brand",the commodity producers,sales or service providers established by the signs for goods and services.Due to the dramatic increasing of the trademark information in today's era,the efficient management and rapid indexing of the rapidly expanding multimedia information became an urgent problem.Owing to the fact that the shape features are the most important features of the trademark image that can describe the object reliably,shape-based trademark image retrieval has been an active research area in image retrieval.The main contributions of this thesis are as follows:(1) We introduced the pretreatment essential method of image.In view of trademark's characteristic,we realized the normalization,the denoising,the binaryzation,the gradation and the image intensification with the procedure.(2) We studied the key features of image-based shape extraction methods.In this paper,image segmentation,contour extraction and border tracking were discussed in detail and realized these methods.In this paper,image segmentation using morphological watershed segmentation method,in view of the watershed over-segmentation of the image of a region we improved the algorithm to merge.Then,we used a improved snakemodel to search the trademark contour.At the border tracking algorithm we used the maximum gradient method.(3) We studied description parameters of shape feature and realized many kinds of description parameters of shape feature,including Shape parameter,invariant moments, border moments and fourier descriptors.(4) We studied the key technologies of the shape-based trademark image retrieval system.Summed up the characteristics shape-based image retrieval system trademark to be involved in key technologies,including database storage,feature matching,feedback and evaluation techniques of search results.(5)We designed the shape-based image retrieval system,which included 3 sub-systems and five modules,users can also have more precise search results based on the results of a feedback.