基于FPGA的高斯建模运动目标检测算法

摘要

The moving object detection technology is to distinguish the moving objects from background in image sequence. It supplies necessary information for advanced motion analysis, hence the rapidity and accuracy in moving object detection take great important part in post processing. In order to reduce the time consumption of object detection and meet the high-speed applications, a moving object detection algorithm with Gauss modeling based on FPGA is designed and implemented on the platform of Cyclone Ⅱ Altera FPGA. The pipeline computational structure and the method of changing floating-point operations are applied to integer ones. The experimental result shows that the design gains good detection results and high precision, and simultaneously runs fast and stable on FPGA.%运动目标检测技术是在图像序列中将运动目标从背景图像中提取出来,为高级运动分析提供必要的信息,因此运动目标的快速正确检测对后期处理至关重要.为了满足高速场合的要求,减少目标检测的消耗时间,采用Altera公司的CycloneII系列的FPGA,研究设计了基于FPGA的高斯建模运动目标检测算法.该算法采用了流水线方法设计算法结构,以及浮点化整的方法处理浮点计算,实验结果表明,该设计在FPGA上运行快速稳定,检测效果良好,并且精度高.