基于FPGA的运动目标实时检测跟踪算法及其实现技术

运动目标检测跟踪有关的算法及其基于PC平台的实现已经比较成熟,但实时性较差。将采集的彩色视频流分成灰度和彩色两个数据流,灰度视频用于目标检测,彩色视频流用于跟踪显示。以经典的帧间差分法和背景差分法为基础,根据现场可编程门阵列(FPGA)的特点及片外同步动态存储器的存取控制要求,对这两个算法用FPGA逻辑单元进行了设计和实现。对原始彩色视频流和转换后的灰度视频流的存取使用乒乓操作,在滤波和形态学处理时使用了并行的流水线操作,极大地提高了算法的实时处理能力。在FPGA开发板上构建了一个彩色视频图像中运动目标检测跟踪系统,对系统性能进行了测试。实验结果表明,系统可在多种分辨率和帧率下进行运动目标进行实时检测跟踪;固定背景差分法对目标运动速度无限制,但当使用帧差法对快速运动目标进行有效的检测时,应使目标的帧差间距大于3.2像素。

Real-time detection and tracking algorithm of moving target based on FPGA and its implementation technology

The algorithms related to moving target detection and tracking and their implementation based on PC platform have been mature, but real time performance of such systems is usually poor. A captured color video stream is divided into two streams: gray and color. The gray video stream is used for target detection, while the color video stream is used for tracking display. Based on the classic interframe difference method and background difference method, and according to the characteristics of Field Programmable Gate Array(FPGA) and the access control requirements for the off-chip synchronous dynamic random access memory, the two algorithms are designed and implemented with FPGA logic units. Ping-pong operations are used for the access of original color video stream and converted gray video stream, and parallel pipeline operations are used for filtering and morphological processing, which greatly improves the real-time processing ability of the FPGA-based algorithm. A detection and tracking system for moving targets in color video is built on an FPGA development board, and the performance of the system is tested. Experimental results show that the system can detect and track moving targets in real time at various resolutions and frame rates;the fixed background difference method has no limit on the target velocity;and however when the frame difference method is used to detect fast moving targets effectively, the interframe difference spacing of the target should be greater than 3.2 pixel.