三维扫描仪中人头轮廓线提取方法及实现

为解决三维扫描仪的实时性,文章提出了以FPGA处理器与PC主机交互式共同完成提取轮廓线的快速算法。该算法由两个阶段组成:第一阶段由主机计算背景与目标的分割阈值。第二阶段由FPGA处理器实时检测轮廓线位置信息。该快速算法具有计算简单、实现速度快等优点,并且减少了传输与存储的数据量,减轻了后面主机计算工作量。同时,省掉了昂贵的图像采集压缩卡与高速硬盘,降低了成本。可重构FPGA处理器设计成流水线结构,对每个像素的平均处理时间控制在70ns以内。仿真与综合结果表明:从一帧720576标准PAL制视频图像中提取轮廓线信息可在40ms内实时完成。

Design and Implementation of the Fast Algorithm for Extracting Contour in 3D Scanner

A fast algorithm of extracting head contour is proposed for real-time image processing,which is accomplished by FPGA processor and host computer together.First,the segmentation threshold between background and head is computed by the host computer.Second,the head contour is detected in real time by FPGA processor.The advantages of the fast algorithm include easy computing,and fast implementation,etc.The algorithm reduces greatly data throughput of transmission and storing and alleviates the computational load of host computer.The devices are low cost by cutting off the expensive image capture and compression card and the high-speed hard disk.The FPGA processor is designed as pipelined architecture.The average time of processing every pixel in each step is less than70ns.The results of emulation and synthesis show that extracting head contour from a frame of standard PAL video image is completed within40ms.