基于FPGA的多运动目标检测算法

Multi-moving Target Detection Algorithm Based on FPGA

针对当前运动目标检测实时性差的问题,设计了基于FPGA的多运动目标检测系统。以帧差法为基础,提取运动目标轮廓,对每个像素进行判断并合并同一目标进行多目标检测与分割。系统主要由图像采集模块、图像处理模块、图像缓存与显示模块组成,CMOS摄像头采集图像数据,SDRAM作为数据流的实时缓存,基于Verilog语言进行算法设计,最后通过VGA进行显示。测试结果表明,系统能快速检测出视频图像中多个运动目标,与PC机相比,在实时性上具有明显优势,且只消耗总体逻辑单元2767个、寄存器1557个。

In view of poor real-time detection of moving targets,a multi-moving target detection system based on FPGA was designed.Based on frame difference method,the contour of the moving target was extracted;each pixel was judged,and the same target was combined for multi-target detection and seg⁃mentation.The system was mainly composed of image acquisition module,image processing module,im⁃age cache,and display module.The image data was collected by CMOS camera;SDRAM was used as real-time cache of data stream;algorithm design was based on Verilog language,and finally VGA was used for display.The experimental results show that the system can quickly detect multiple moving tar⁃gets in video images,which has obvious advantages in real-time performance compared with PC and on⁃ly consumes 2767 total logic units and 1557 registers.