Image Super Resolution And Target Detection Via Parallel Computing Based On Multi-core CPU

With the continuous development of imaging techniques, the amount of data within digital image has become larger and larger.In order to acquire higher and higher precision at a variety of image processing results, the variety of new image processing algorithms have been proposed.But their computational workload becomes heavier and heavier.Especially when dealing with large-scale images,sequential algorithms can hardly satisfy the application requirements due to their low processing speed. It is an effective solution to accelerate image processing speed by using parallel computing technology.Now multi-core CPU processors are widely used.Parallel algorithms based on multi-core CPU processor have good adaptability,which are used in most environments.Parallel computing base on Multi-core CPU can take full advantage of processor resources and improve resource utilization. In this paper,we design the super-resolution parallel algorithm and aircraft detection parallel method to speed up processing,which is based on multi-core CPU parallel computing technology.The main research work is described as follows:(1)Based on the platform of MATLAB parallel computing, we propose the parallel algorithm of image super-resolution based on the compressed sensing.The sequential algorithm has good effect of super-resolution reconstruction.But time complexity of the sequential algorithm is too high and its process speed is too slow. This paper describes the basic principle of the sequential algorithm and its implementation steps.By analyzing time complexity of the sequential algorithm and making experimental verification, we found that there is a computationally intensive iterative process which is the key part of affecting processing speed.In this paper,using the MATLAB parallel toolbox,we propose a parallel algorithm.Many small tasks which the time-consuming task is decomposed into are computed synchronously in many parallel processing computing units.The experimental results show that the parallel algorithm can greatly reduces the time complexity and speed up the image processing when maintaining the performance of the sequential algorithm.(2)We propose the parallel scheme of the most time-consuming part of the aircraft detection method based on regional connectivity.In this paper,we describe aircraft detection steps in detail.The sequential method has high time complexity.we found that the most time-consuming part is the image pre-processing stage.So we design a parallel scheme by parallel programming technique based on Open MP for the part.The most time-consuming part of the computing task is decomposed into fragments which are computed synchronously in multi-threaded running on multi-core processors.The parallel scheme makes full use of the processor resources to achieve the effect of the acceleration.The experimental results show that the parallel method achieves good effect of acceleration on the part and reduces the aircraft detection time.