Real-time Image Fusion Of Double Dsp Parallel System Research

With the development of science and technology, the performance of DSP has been improved greatly and has been used widely in image processing field. In order to realize the real-time processing of image with high-number frames and big size, as well as complex algorithms, a real-time image processing system requires a large amount of data processing capability to keep the real-time nature. Due to the frequency limitation, an image fusion system with single DSP core cannot meet the requirements of large amount scientific computing and high speed real-time signal processing. To solve this problem, designing and developing a parallel-working processing system with strong real-time nature, high precision and enhanced data throughput is an effective method to increase the overall data processing ability.In this paper, we design and develop a parallel-working real-time image fusion system based on the existing image fusion system with single DSP, in which dual-DSPs are the processing core and FPGA is the control core. Firstly, introduce the hardware structure design of this real-time image fusion system, including the overall design, as well as the design of different function modules; Secondly, introduce the design of software and driver developing, including the integrated development environment CCS setting in the process of debugging the dual-DSP system and the video capture driver development; Then, analysis the general method to evaluate the performance of dual-DSPs system; After that, introduce the algorithms design in a dual-DSP system, including implementing the Laplacian pyramid algorithm on the hardware system and transplanting the existing algorithms in single DSP system to the parallel-working DSP system; Finally, through the experiments and the analysis of experimental data we make a conclusion that design a dual-DSPs parallel-working system can improve the data processing capability greatly.