Ant Algorithm Of Cluster Research And Its Application In Texture Image Processing

The main human visual perception of objects from color, texture and shape. Texture as asurface, a basic inherent properties exist widely in the nature, is described and the identity of theobjects of a kind of very important feature Texture analysis is an important goal of human visualdistinction between one of perception function. Machine vision of the main purpose of the study istexture understanding, modeling, processing in the images of the texture mode, finally usingcomputer technology to texture of human visual simulation study and cognitive processes Takingthe theory of constructivism as the theory base.Ant colony algorithm is a kind of new bionic algorithm, the algorithm was originally used tosolve traveling salesman problem (TSP), then with the development of the computer science, arebeing used for other dynamic optimization areas. The algorithm has the positive feedback,distributed computing, robustness and other characteristics, and simple structure, easy to berealized, easy to realize the combination with other algorithms. Due to the ant colony algorithmhas a strong clustering ability, good at dealing with discrete digital texture image. This subject willtry ant colony algorithm into texture image processing, including image segmentation, match theimage recognition.This article mainly includes the following four aspects:(1)the ant colony optimization (ACO) the emergence, development and basic principle, thebehavior of ant colony algorithm model and the convergence of ant colony optimization algorithmand its simulation.(2)use of interest is designed on the basis of pheromone ant colony algorithm of imagetexture segmentation algorithm. Algorithm is mainly through the statistical characteristics ofregional texture information, such as computing grayscale information (or color information), fieldof similarity information space position, use pheromone and choose interest preference calculatingbrought between the region transition probability, using the transition probability with small area.The experimental results show that the algorithm is feasible, using the interest pheromone reallycan accurately reflect the image texture model.(3)in the basic ant of cluster model and quad tree based on the initial segmented, from imagetexture feature, field similarity distance, probability conversion functions, three aspects oftraditional ant of cluster improved algorithm, and puts forward a new cluster of ant algorithm-fourbinary tree based on ant colony algorithm (QTACA), and in the same experiment environmentmore texture image of simulation experiment research, the results show the modified algorithm forafter the optimal solution of the at a significantly faster, better segmentation effect.(4)use of color and the entropy of the ant algorithm of image of cluster match, the extractionof color image for registration and entropy information, will each characteristics as little ants,clustering center is food source. An ant calculation of information on the path to stay at each meal,can set, every main path of pheromone strength for the bottom of the path length, and finally topheromones of the maximum sum for the best matching. Through the image color features andentropy characteristics, in the database looking for matching images...