Circuit Design of Moving Object Recognition System

Abstract

The traditional moving object recognition and tracking systems are usually based on software environment of PC, so the performance of processing speed, real-time and the size have limitations. In view of this, the circuit design based on Field Programmable Gate Array(FPGA) is proposed, which improves the performance of processing speed and real-time and miniaturization. Aiming at the recognition, tracking and feature point extraction of moving object, 300,000 pixel camera is used for image acquisition, and hardware circuit is used for image processing, including real-time image caching, gray-scale processing, improved bit plane median filtering, segmentation of self-adaptive threshold binarizing, fusing of frame difference and background difference methods for moving object detecting, tracking and feature point extracting. Finally, the recognition results are displayed by HDMI interface displayer. The design takes full advantage of the high-speed and parallel processing ability of FPGA, and combines the high-speed on-chip RAM with the large capacity off-chip SDRAM to realize the processing and storage of video data. After written by Verilog HDL and verified by Modelsim, the physical circuit is implemented in FPGA. This system has the characteristics of strong anti-interference, compact, flexibility, high speed, low power consumption, versatility and scalability, which is suitable for both industrial field and home use.