Shape Classification And Its Application In Image Retrieval System Research

All the countries have taking up with the construction of the modernization ofthe education. And the construction of the education resources repository is the basisof the informatization of the educational. It is a project that needs to be constructedand maintained secularly. The effective use of the education resources repositoryplays more and more important part in the development of the education.Advances in computing and networking are generating a significant amount ofdigital images and the images are important resource of the education resourcesrepository. But how we can find the image we want quickly and efficiently? It becomea problem that needs to be desiderated.To copy with this problem, there have been two major approaches which studyimage retrieval from somewhat different angles. The traditional approach is toannotate each image manually with text which describes the content of the image;some image attributes such as number and date can also be included in the annotation.Another approach is to index images directly based on the visual image content.There are several problems with the text-based image retrieval. To overcome theproblems, content-base image retrieval (CBIR) was proposed in the early 90's and ithas been under intensive research from then on. Content-based image retrieval (CBIR)system is designed to help retrieve relevant images in an image database based ontheir image contents which include their visual and semantic content, for example,color, shape, and texture. Many image retrieval systems, both commercial andresearch, have been built, for example, QBIC , Photobook, VisualSEEk and so on.Shape is one of the primary low-level image features exploited in CBIR.Although color and texture contain important information, if two images with similarcolor histograms, they can represent very different things. Therefore, the use ofshape-describing features is essential in an efficient content-based image retrievalsystem.In this thesis, we firstly make a review of the shape description and then we usethe different shape features to discriminate the shape.During the research, we developed a system based on shape recognition and animage retrieval system with VC++ and Matlab based on Win2000 operating systemand the system achieves both the desired efficiency and accuracy.