The Research And Application On Swarm Intelligence Algorithm Of Image Compression

The image compression is under the requirement condition and in the certain guarantee of image quality, reduces the primitive image data quantity, which may sum up a complex optimized problem.The evolution algorithm is the effective method of soluting complex optimization; it belongs to the evolution computation application domain.Therefore, this article will study evolves algorithm in solving the image compression problem.This method not only makes the algorithm operation to be simple, but also data processing to CPU and memory request is not high - with extremely few bit number memory image data, enhances the image the quality.The article utilizes three kinds of Swarm Intelligence Algorithms- Genetic Algorithm (GA), PSO (Particle Swarm Optimization) as well as Quantum-behaved Particle Swarm Optimization (QPSO) to study the image compression problem. Considering the image compression sums up as the error function minimum to memory image bit, but the error function generally has multi-peaks, and possibly produces many local values, additionally the general gradient algorithm often cannot find the global convergence optimal solution. PSO algorithm and the QPSO algorithm are the global convergence algorithm, therefore optimizes the data reserves with the PSO algorithm or the QPSO algorithm regarding to enhance the picture the quality to have the vital significance.The paper firstly studies genetic cluster method based on the image pixel in order to obtain to an ordered representation of the image and then perform the clustering (such as fixed clustering, dynamic clustering with fixed clusters, dynamic clustering with variable clusters, local tuning) to obtain the compression result, then carried on the experiment to analyze its feasibility and four kinds of algorithms comparison. The simulation result indicated that, gathers kind of GA with the local tuning to carry on the image compression, the precision high, and the convergence speed is quicker.Next it was studied the feasibility of applied Particle Swarm Optimization algorithm to image compression processing, simultaneously in order to enhance the convergence rate of the algorithm, enhancement algorithm global search ability, a kind of Quantum-behaved Particle Swarm Optimization algorithms model QPSO was proposed. It mainly modify the quantum physics thought to revise PSO "the evolution" the method (namely renewal particle position method), when particle position was updated, each particle with emphasis the current partially best positional information and best positional information of the global situation should be considered. Lastly, comparison with QPSO algorithm and genetic algorithm, it was discovered that the restrain in the effect in the time of compression to surpass the genetic algorithm.