Real-time Recognition Of Mark Point Coordination Based On FPGA

Videogrammetry Measurement(VM)is widely used in aerospace,biomedical science,remote sensing telemetry,military and so on.VM needs to place artificial mark points on the surface of measured object,but the existing mark point image recognition process is more time-consuming by using computer.For the high frame frequency/high resolution camera,its mark points' coordinates cannot acquire real-time.Therefore,this paper focuses on FPGA-based image coordinates real-time identification techniques for VM mark points.The algorithm of coordinate recognition is studied,which is used to acquire the image coordinates of VM mark points(the test image resolution is 2352 × 1728,its frame rate is 60fps),and the test results show this algorithm is effect and robust by using Matlab simulation.A fixed-point multiplier based on Vedic algorithm is designed for a given specific hardware,and its verification experiment show that the 16×8-bit wide-multiplier designed by this is fast 10 times as that of the 4-Booth algorithm multiplier,and it consumes less resources than the 4-Booth algorithm multiplier.Using the "top-down" design method based on hardware-oriented,the image coordinates real-time identification for VM mark points is divided into different modules in Verilog HDL language,and the FPGA parallel pipeline algorithm is employed to achieve the hardware acceleration.ISE,Modelsim and other EDA tools are used to simulate,verify and do timing analysis for these various functional modules.The FPGA platform of image coordinates identification for VM mark points is established,its experimental results show the occupied resources no more than 60%,and the acquired coordinates of the mark points are correct compared with the existed VM software results.