基于计算机视觉平面3D效果显示.docVIP

基于计算机视觉平面3D效果显示 　　摘 要： 3D显示技术作为当前计算机图形图像技术的研究热点，以其在平面上实现真实的视觉体验而获得高度重视。为了提升用户的3D体验，提出了一种基于计算机视觉的3D显示器。借助于Windows操作系统，通过OpenCV操作红外图像采集设备，搭配可见光滤光镜，消除了人眼反射光的干扰。提出了一种比较高效的人类瞳孔定位算法，首先利用灰度积分投影将人眼的大概位置确定，之后使用优化的阈值分割方法得到二值图像，在此基础上对其进行形态学运算去除噪声，使图像更易处理。使用基于OpenCV的角点检测与椭圆拟合结合的方法定位瞳孔，算法简单高效；基于OpenCV的cvCalibrateCamera函数实现摄像机定位；最后使用Bresenham画线法完成了3D显示的设想图。为了使系统拥有更好的实时性，对系统使用了分割扫描区域、预测目标区域两种方法降低系统运算量，实现了人眼的快速追踪。 　　关键词： 计算机视觉； 3D显示； 人眼检测； 数字图像处理； 阈值分割 　　中图分类号： TN911.73?34； TM417 文献标识码： A 文章编号： 1004?373X（2016）11?0070?04 　　Abstract： The 3D display technology is the research hotpot of current computer graphic image technology， and high attention is paid to realization of its true vision experience on the plane. In order to promote the 3D experience of users， a 3D display based on computer vision is presented， which can eliminate the reflected light interference from human eyes by means of Windows operation system， infrared image acquisition device operated by OpenCV， and visible light filter. An efficient human pupil location algorithm is proposed， by which the approximate location of human eyes is determined through gray integral projection， and then the binary image is got with the optimized threshold segmentation method. On this basis， the morphological operation for the binary image is conducted to eliminate the noise and make the image processing easier. The algorithm combined corner detection based on OpenCV with ellipse fitting is used to locate the pupil， which is simple and efficient. The camera location is realized on the basis of cvCalibrateCamera function of OpenCV. Finally， the envisaged image of 3D display was accomplished with Bresenham line drawing method. To make the system have better real?time performance， the scanning area segmentation method and target area prediction method are used to reduce the system computing amount and realize fast tracking of human eyes. 　　Keywords： computer vision； 3D display； human eye detection； digital image processing； threshold