APP下载

一种基于计算机视觉的手指特征识别算法研究及其FPGA实现

2017-06-23冯清娟何志宽

现代电子技术 2017年12期

冯清娟+何志宽

摘 要: 为了提高视频图像中手指特征识别的精度和速度,提出一种基于矩形模板的动态阈值算法。该算法采用矩形模板平滑图像,用灰度阈值法提取目标边缘,实现对手指特征的准确识别。Matlab仿真结果表明,该算法受环境噪声影响较小,能清晰准确地分割出手指边缘。在FPGA平台上设计了一套基于计算机视觉的手指特征识别系统,并对该系统的实时性、手指识别精度等性能进行了测试。测试结果表明,该系统的运行速度与摄像头摄取图像的速度基本同步,可满足系统对实时性要求。系统对手指位置识别的坐标偏差约为3个像素,基本满足系统对识别精度的要求。

关键词: 计算机视觉; 手指特征识别; 动态阈值算法; 矩形模版

中图分类号: TN911?34; TP37 文献标识码: A 文章编号: 1004?373X(2017)12?0113?04

Abstract: In order to improve the accuracy and speed of the finger feature recognition in video image, a dynamic threshold algorithm based on rectangle template is proposed. The rectangle template is used in the algorithm to smooth the image, and the gray threshold method is used to extract the edge of the target to recognize the finger feature accurately. The Matlab simulation results show that the algorithm is lightly influenced by environmental noise, and can segment the edge of the finger clearly and accurately. A finger feature recognition system based on computer vision was designed on FPGA platform. Its real?time performance, finger recognition accuracy and other performances were tested. The test results show that the running speed of the system is synchronous with the speed of capturing the image by camera, which can meet the real?time requirement of the system. The coordinates deviation of the finger position recognized by the system is about 3 pixels, which can basically meets the accuracy requirement of the system.

Keywords: computer vision; finger feature recognition; dynamic threshold algorithm; rectangle template

隨着科学技术的飞速发展,人与计算机之间的交互活动越来越密切,并逐渐成为人们日常生活的重要组成部分。传统触摸屏是现阶段一种最常见,也是最重要的人机交互方式,其是通过一种附加在显示器表面的透明介质,依赖使用者的手指触摸该介质来实现对计算机的输入控制[1?2]。目前,这种触摸屏技术已经相当成熟,但是,技术上不容易做到大尺寸,且成本相对较高,也易于破损。为了突破传统触摸屏这种人机交互方式的局限,把任何成像平面变成可用手指进行交互的系统,本文提出了一种基于计算机视觉的手指特征识别算法,并在FPGA平台上建立了一套基于计算机视觉的手指特征识别硬件系统。……

登录APP查看全文