Partition-Based Image Classification And Retrieval Using SVM

Modern techniques produce all kinds of information in different ways. Image is the most important multimedia information,and the amount of images is growing fast.Therefore, it becomes an urgent problem that how to find needed image efficiently in large-scale image database.Two effective ways has been proposed to solve the problem : one is content-based image retrieval(CBIR) technique which search target images by low-level content feature. The other is semantic image retrieval technique which in search of ways to get images'semantic information.In this paper, machine learning method is used to learn image features and to automatically construct models for image classes. Support vector machines(SVM) are trained for nature image classification and retrieval,thus providing users with a conceptualized way to image query.In the approach of content-based image retrieval, there exists a semantic gap between low-level visual features and high-level concepts. This paper proposes an image representation method based on image partition and region clustering. We try to combine spatial information with color features.This is achieved by partitioning images in the training set into fixed size cells,and for each cell ,extracting a local color histogram as the color invariant feature of the cell.All of the color invariant features are clustered into a number of patterns. The mapping which maps an image to its representation does not really depend on that image alone but on the entire collection of images from which the region groups have been built. SVM are trained to learn the model of the image classifications,and this is a efficient way to solve the problem of semantic gap between low-level visual features and high-level concepts.