Image Object Segmentation By Combining Bag-of-words Models And Context Information

Image object segment, to segment the specific object from clustered background, is the ultimate goal of image segmentation, is image segmentation combined with object recognition. Image segmentation technique has been widely used in almost every field related to image, and image segmentation is directly related to image analysis and image understanding , related to whether we satisfactorily complete visual task or not. So image object segmentation is a worthy field to investigate.This paper addresses the problem of accurately segmenting instances of object classes in images without any human interaction. Our model combines a bag-of-words recognition component with spatial regularization based on a random field, a Dirichlet process and a Gibbs sampling mixture. Bag-of-words recognition component is supported by image feature extraction and visual vocabulary building. Bag-of-words models successfully predict the presence of an object within an image; however, they can not accurately locate object boundaries. Random Fields take into account the spatial layout of images and provide local spatial regularization. Yet, as they use local coupling between image labels, they fail to capture larger scale structures needed for object recognition. These components are combined with a Dirichlet process mixture. It models images as a composition of regions, each representing a single object instance. Gibbs sampling is used for parameter estimations and object segmentation.First, we put forward a kind of SIFT dense sampling algorithms, and use this method to extract image features. Second, we implement an effective algorithm of constructing visual vocabulary, and Image is described with visual vocabulary. Third, we construct object segmentation model that is composed by a random field, a Dirichlet process and a Gibbs sampling, and complete image object segmentation system. Last, we present results on the TU Graz-02 database. The TU Graz-02 set contains images of the categories bicycle, car and person.The experimental results verify the capability of our image object segmentation system. Our model successfully segments object category instances, despite cluttered backgrounds and large variations in appearance and viewpoints.