Research And Application Of The Embedded Image Parallel Processing System

At present, the work of processing image information is often based on PC. However,PC is bulky and inconvenient to carry, meanwhile, in the process of using a lot of supporting electric power facilities, environment and other resources is strictly required. Consequently, real-time processing of image data has certain difficulty in the wild. Thus, this thesis is going to make research on the embedded parallel DSPs processing system to realize the real-time processing of image data.For processing data information of the image, huge amount of data is needed to be dealed, meeting the real-time, consequently, the requirement on the speed of processing platform is higher. This article chooses high-powered performance DSP chip, which improves the processing speed of the platform. The whole process of hardware platform is to use camera to collect simulated image of the outside world, convert analog signal into digital information through the analog-to-digital conversion, preprocess of digital information and store the result which is transferred by the serial RapidIO bus to the parallel processing platform and processed by the parallel processing platform, transfer the processing result to the user at the end.Since dual-DSP processing platform needs to meet real-time, how to rapidly transfer the precessed data is a problem. This thesis selects the serial RapidIO bus to meet the system’s real-time. The serial RapidIO bus connects between the DSP and DSP, DSP and FPGA, making the maximum data transfer rate up to3.125Gbps and letting data transmission without delay come true, that solve the problem of fast transferring mass data and meet the real-time requirements of the system.Software development is done by the integrated development environment CCS provided by Texas instruments.The simulation of DSP hardware environment in CCS is used to design.First, the algorithm function is realized on PC, then transplanted and optimized to let it run on parallel platform. It uses genetic algorithm to solve parallel experiments of TSP problem and verify the advantages of parallel processing platform.At last, the dual-DSP parallel processing platform proposed in this thesis, is used to process the image data of field drains. Experiments show that traditional computer processing mode can be replaced by the embedded parallel image processing system, which relatively has a higher processing speed at the same time.