Adaptive Image Retrieval Based On Genetic Algorithm

As an important kind of information, images have been used widely. So a problem is put in front of us: how to retrieve useful information from so much images?Genetic algorithm, as a machine study method, is greatly enlightened by biological evolution. It has been used in image retrieval and has more advantages than other machine study methods. However, there are only a few researches on genetic algorithm in image retrieval. There are following two reasons: (1) It is difficult to hypothesis because there are so many image features as to denote a whole image. (2) Genetic operator makes little sense to the genetic operation. To result the above two problems, this paper bring forward a few retrieval model based on genetic algorithm, The main research work as fowling:(1) Profound research on the image retrieval based on color and texture feature. The paper analyses the advantages and disadvantages of the two methods. Then each image in the collection is segmented into a constant number of sub-images, and the color and texture feature content in each sub-image is computed to make up the feature vector of the image.(2) Discuss other feedback mechanism in image retrieval and analysis each disadvantages and advantages. It particularly describes the implement procession of the genetic algorithm, expand the influence of the fitness function and genetic operator to the retrieval performance.(3) Bring forward a new genetic algorithm, which is a self-adapting retrieval strategy. The feature vector of every image is reshuffled in a helix way to produce the image chromosome, which is used in the following genetic operations. According to the distance between the example image and the first population, put forward a new adaptability function and genetic operator. Elitist strategy is used during the genetic operation to make sure the user does not lose the fittest result. After intercross, variation operation, produce the next chromosome and then find out the corresponding image in the image library to show.(4) Design an experiment system to improve the efficiency of the genetic algorithm by compare to the global color histogram and feedback method.This paper prove the genetic algorithm is not only practically and theortically efficient, but also can make the retrieval system more intelligent.