A High Performance SURF Image Feature Detecting System Based on ZYNQ

Abstract

In this paper, a power-efficient and real-time image feature detecting system is implemented, which is based on the Speeded-Up Robust Feature (SURF) algorithm. We optimized the SURF algorithm, and implemented on the FPGA fabric of Xilinx ZYNQ-7020 device. Our design of SURF algorithm circuit can work up to 100Mhz clock frequency, and its processing speed up to 270 fps for standard VGA (640 * 480) resolution gray image. We implemented the system on the ZYNQ platform with the hardware and software co-design approach. The image feature detecting system based on SURF algorithm circuit runs embedded Linux system. There is a GUI application for Linux system designed with QT and open-cv, which can capture video, process and display image or video. The system meets the real-time and low-power requirements of embedded devices, with great practical value.