Study And Implementation Of The Real-time Image Rotation System Based On FPGA

At present, the major problem in image processing is the high requirements on processingspeed. As the basic transformation, image rotation becomes to be the bottleneck of thereal-time features of the whole image processing system, because it needs to handle largeamounts of data, and involves the non-sequential memory access. Comparing with thecomputer and DSP, FPGA is programmable, reconfigurable and has the feature of parallelcomputing, which makes it suitable for high-speed digital signal processing. Therefore, thispaper studies on the real-time image rotation system based on FPGA.To realize the real-time image rotation, processing capacity should be improved. Thereare three methods. Firstly, parallel computing, which means that two or more processor areused; secondly, optimizing the algorithm to get higher efficiency; thirdly, improves the systemfrequency and chooses the better memory access scheme. In this paper, the followingmeasures are adopted to make the methods to be practice: using FPGA as the video streamprocessor; choosing the direct method which is suitable for FPGA to realize the coordinatetransformation after analysis and comparison. And furthermore, in order to meet therequirements of sequential and non-sequential access, both SDRAM and SSRAM are used.During the design, the design methods of high-speed digital circuit like ping-pong operation,time-sharing reuse, pipeline operation are took to improve the processing speed of imagerotation with the full consideration of the characteristics of image processing based-on FPGA.The system is designed in QuartusⅡ platform, combined with hierarchical and modularthoughts, and programmed with VerilogHDL. Meanwhile, the timing closure is achievedusing TimeQuset tool, thus computing power and stability of the system is improved. Finally,the system is verified on the common image processing platform, and its processing capabilitycan meet the processing speed requirements for the real-time image processing system.