For the image processing, computer vision and 3D rendering have the feature of massive parallel processing, the programmability and the flexible mode of parallel processing on the Polymorphic Array Architecture for Graphics (PAAG) platform were utilized adequately, the parallelism design method by combing the operation level parallelism with data level parallelism was used to implement the OpenVX Kernel functions and 3D rendering pipelines. The experimental results indicate that in the parallel implementation of image processing of OpenVX Kernel functions and graphics rendering, using Multiple Instruction Multiple Data (MIMD) of PAAG in parallel processing can obtain a linear speedup that the slope equals to 1, which achieves higher efficiency than the slope as nonlinear speedup that less than 1 of Single Instruction Multiple Data (SIMD) in traditional parallel processing of the Graphics Processing Unit (GPU).

When Artificial Bee Colony (ABC) algorithm is applied to the load balancing system, local intensive resources may arise due to improper handling of the relationship between requests. Therefore, an improved adaptive ABC algorithm was proposed. By introducing other types of requests, the request accumulation of the similar type got blocked. Then, not only the availability but also the system throughput had been improved. The experimental results show that either the system throughput or the scalability of the load balancing mechanism based on ABC algorithm, is applicable to cloud computing environments. The improved adaptive ABC algorithm is superior to traditional algorithms more than 5% in system throughput.

It plays an important role in realizing cloud computing to implement the load balance of accessing resources. Therefore, based on the characteristics of cloud computing environment, an improved fuzzy clustering analysis algorithm was proposed in this paper. Furthermore, integrating Particle Swarm Optimization (PSO) algorithm with fuzzy C-means algorithm improved the algorithm accuracy. Then, using the improved fuzzy clustering algorithm in analyzing utilization ratio of I/O and Central Processing Unit (CPU) of all computing nodes, each node was divided into an affirmatory collection. And it represented its load level, as a basis for judging the node which needed to transfer tasks, so as to achieve the load balance. According to the experimental results, in terms of algorithm accuracy, no matter what is in the UCI machine learning repository, or for the proposed load balancing mechanism, the improved fuzzy clustering algorithm's reached the traditional one's by 110%. Besides, it surpasses the traditional one in the stability of the algorithm.