Review of the sparse coding and the applications on image retrieval

Image retrieval is based on the description of image content, found in the target image collection with the specified characteristics or with the image of the specified content. The content of the image can be divided into two categories: visual content and information content. The visual content corresponding to the physical representation of the image, such as color, shape, texture, etc. while corresponding image semantic information content, such as the theme, characters, scenes, etc come under the category of information content. The emergence of the theory of sparse representation for computer vision and pattern recognition, and other fields has brought the profound influence. Image sparse representation is a well as a new image description model, the said coefficient of less than zero composition reveals the image signal of the main structure and essential attribute. Efficient representation of the model not only can obtain the observation signal, but also can extract the semantic information. This Topic through the introduction of sparse coding idea retrieval process can be divided into two steps: off-line training and on-line retrieval. By using the sparse dictionary online training algorithm for each image to complete dictionary training, using the image similarity evaluation algorithm based on sparse feature expression of query image and the database image similarity calculation, finally according to the similarity retrieval results, thus the ascension of systemic algorithm accuracy and efficiency.