Research On Biologically Inspired Computing Method Based On GPU Acceleration

A lot of complex optimization problems are produced in people's daily production and life.These complex optimization problems are difficult to solve in a conventional way within a reasonable time.In nature,biological behavior tends to reproduce and evolve in favor of the self or the population.With the deepening of research on biologically inspired computing methods,swarm intelligence algorithm has been widely used in the field of optimization as an important computing method.As the application scenario expands and the complexity of the problem increases,if the swarm intelligence algorithm uses large-scale populations for optimization,there will be problems of low efficiency and time-consuming.In recent years,with the rapid development of computer graphics hardware,due to its high-speed data processing capabilities and gradually mature software programming system,it has gained gradually increasing applications in the field of general computing.In view of the above problems,the inherent parallel characteristics of the bionic optimization algorithm can be studied based on the GPU from the perspective of heterogeneous computing,so as to optimize and accelerate the algorithm.The thesis proposes a parallelization design method of swarm intelligence optimization algorithm by analyzing the parallel design scheme of swarm intelligence algorithm under different granularities,and solves the key problems involved.Furthermore,parallel particle swarm optimization,parallel artificial bee colony algorithm and parallel artificial fish swarm algorithm are designed and implemented.Experiments show that the parallel algorithm designed in this paper can effectively improve the execution efficiency of swarm intelligence algorithm,and proves the effectiveness and efficiency of the proposed method.Image segmentation is one of the key technologies in the field of digital image processing and computer vision.The threshold method is widely used due to its simple implementation and stable performance.Multi-threshold segmentation can obtain richer image information than single-threshold segmentation.Due to the use of swarm intelligence algorithm for multi-threshold image segmentation,there is a problem of excessive complexity.Combined with the above-mentioned parallelization method for swarm intelligence algorithm,this paper studies the parallel swarm intelligence algorithm based on GPU acceleration for image segmentation to improve the efficiency of the algorithm.Experiments based on parallel particle swarm optimization algorithm show that the efficiency of the algorithm is significantly improved and a better segmentation result is achieved.