The Design And Implementation Of Embedded Real-time Image Processing Platform Based On ARM、FPGA And DSP

The traditional image processing systems always use DSP as core processor and useFPGA to realize other functions. These systems may have a good real-time performance,while they are poor in system control, interface design, man-machine interaction and etc.In this paper, an embedded real-time platform based on ARM+FPGA+DSP is proposed,which utilizes the excellent control ability and man-machine interaction performance ofARM. The platform has flexible hardware architecture and can be used in many areas suchas image acquiring, automatic target recognition, video monitoring and etc.System control is independent with data computation in the design of the platform.The control part based on ARM contains various peripheral interfaces and enough storingresource. Besides, it can run embedded operation system to manage all the tasks. Thecomputation part is based on DSP and FPGA. DSP is powerful in digital signal processingand FPGA has excellent designable flexibility and parallel computational ability, whichenhance the performance of both control and computation of the platform. What’s more,the platform can be extended to provide more computational resource for the requirementof complicated processing in future.The logic design of FPGA and the interface design of HPI are important and will bedescribed detailedly in the paper. FPGA not only controls data path but also can do someimage pre-processing works. Otherwise, DSP can be mastered by FPGA through HPIinterface.At the end of the paper, a summary of the hardware design is submitted and sometraditional methods have been improved.