Based On Artificial Immune System’s Natural Image Classification Algorithm And Technology Research

The rapid development of network and multimedia technology provides great convenience for digital image storage and dissemination of massive image data can not be completed timely filed only rely on the human, organization and management. How to take advantage of such a large number of image information and to locate the material of interest, a major challenge to the query image information technology. As a new branch of computational intelligence, AIS algorithm is not a generic algorithm framework, and many of the excellent features of the immune system have yet to be applied to the design and application of the AIS algorithm, artificial immune algorithm development and application of space is quite broad. The papers draw on the usage of AIS in other areas, and constantly improve its image classification.First, the paper in the theory and algorithms based on natural image classification algorithm based on artificial immune system. Designed for natural image classification using artificial immune algorithm, you first need to be solved before the abstract as it can handle the antigen in the form. Antigen capture and affinity metrics. Antigen stimulation and identify strategies and clone selection and mutation analysis, Finally, a numerical example, select the image library example environment by1000natural images, including African life, cars, dinosaurs, flowers, horse, landscape, food, motorcycles, aircraft, persons face the10categories of subject matter. Through the analysis of the block weights and feature weights, proved that the algorithm has the highest average accuracy rate of88.68%, and the classification results are satisfactory.Secondly, the paper presents the natural image classification techniques based on color distribution entropy. First, the paper analyzes the color spatial distribution entropy, followed by analysis of the image information entropy, including the color of the spatial distribution of entropy, the spatial distribution of the weighted entropy and natural image classification algorithm based on color distribution entropy. For effective use of the two features for image classification, the weighted distance method can be used in the metric vector similarity. Finally the technology experimental analysis, experiments show, the characteristics at the same time show consideration for the overall characteristics and the distribution characteristics of image color histogram and I-CDE has better classification performance than global. And compared to the I-CDE, the computational complexity of the method lower, time significantly saving feature extraction.AIS as an intelligent computing method based on the biological immune system, provides a novel method and means for classification. On this basis, the paper introduces the value of the right to block in order to eliminate the effects of entropy symmetry. The experiments show that, combined the distribution entropy with global color histogram as the feature vector image classification, classification accuracy improved.