Research On Sopc Design Technology Of Target Matching Algorithm

In recent years,with the rapid development of science and technology,image matching technology has gradually become an important research hotspot in the field of scientific research,and is becoming more and more important in fields of computer vision,medical image,military automatic target recognition,satellite image compilation and analysis.SIFT algorithm is a kind of feature matching algorithm based on scale space,which preserves the invariance of image scaling,rotation and even affine transformation,so its algorithm and its derived algorithm have important significance and value in image matching technology.Nowadays,most of the image matchings are implemented based on software.Because of the huge amount of data calculation,real-time and high accuracy of image matching system is difficult to ensure through software.With the development of FPGA and SoPC technology,the hardware system gradually has enough resources to complete the image processing calculation.At the same time,the processing speed of the image matching can be greatly improved by the parallel processing of FPGA.In this paper,the image matching SoPC system based on SIFT algorithm is researched and designed.The Euclidean distance matching module in SIFT algorithm is researched,including exhaustive method and 2-nearest neighbor algorithm.In this paper,The descriptor data is first stored in DDR,and then the method of" calculation while transmission" are used to realize the hardware migration of the algorithm.At the same time,we use multi-level Pipelining to accelerate data processing,to reduce the use of logical resources,and to optimize timing.Then,the Euclidean distance matching module was modified and improved by enlarging the input data width,increasing the processing module,etc.Meanwhile the processing speed of the module is speeded up.Then,based on AXI bus protocol,the module is encapsulated into Match self-encapsulated IP core,and hardware environment is built based on Zynq-7000 platform,and the hardware part of image matching system is implemented.After that,referring to the software system design flow based on Xilinx SDK,the operating system layer of image matching system is designed and implemented by C language to realize the data transfer and parameter control between ARM and FPGA.After simulating the Euclidean distance matching module with Software Modelsim,The entire image matching system is tested based on hardware and software co-test technology by using the ZC702 development board.The final test results show that the image matching system not only meets the design requirements,but also has good real-time.