Based On More Than Soft-core Parallel Processing, Digital Image Processing Algorithms

Digital image processing is a technology using the computer algorithms to perform image processing on digital images.The upper computer software programming is most commonly used to perform digital image processing, which is hard to gratify the demand of the real-time image. Using hardware to perform the digital image processing will be the trend.FPGA is particularly suited for the digital image processing for its ample logical resources. However, the parallel processing using the multi-soft core technology can enormously quicken the image processing speed, gratifying the demand of the real-time image. FPGA has many advantages:it can achieve greater design flexibility, greater program and module portability; it can shorten the design circle and reduce the hardware investment risks. Therefore, FPGA will be widely used in image processing field.System Generator is a system modeling tool of Xilinx, extending the MATLAB Simulink platform in many ways, providing the DSP modeling environment suitable for the hardware design, accelerating and simplifying the DSP system hardware design of FPGA. System Generator can provide system design capability, allowing the hardware simulation, implementation and verification under the same circumstances, without writing the HDL code. Besides, System Generator enables high-level abstraction and automatically generates FPGA code. Moreover, it enables low-level abstraction and visits the infrastructure resources, fulfilling the high-efficient building of FPGA. By far, the design method based on System Generator has already demonstrated its great power in complicated system realization and it will surely become one of the mainstream FPGA development techniques in the future.This article introduces several commonly-used algorithms in digital image processing and makes modeling and simulation analysis of DCT in digital image processing algorithms. This model uses the FPGA multi-soft-core technology and the two EDA tools of Xilinx System Generator/ISE Design Suite10.1 and Mathworks Simulink/MATLAB R2007b, and uses the multi-DSP modules and adopts parallel processing. The result of the model has already been verified in MATLAB and its validity has been proved.