增强现实是一种虚实信息融合的技术。随着智能手机的普及,移动平台增强现实的应用开发成了热点。在智能手机硬件计算资源有限的情况下,基于人工标识的移动增强现实系统存在标识检测识别算法计算量大、配准跟踪实时性较差等缺陷。针对上述不足,提出了利用水漫填充法和外接矩形的几何特征快速检测标识4个角点,设计了基于海明编码特征的人工标识,提高了标识的识别速度,并具有一定的抗局部遮挡性能。为了精确快速地求解摄像机位姿,系统采用Levenberg-Marquardt算法优化求出的平面单应性矩阵。最终根据优化的单应性矩阵求解出摄像机位姿,并根据求出的位姿对虚拟物体加载渲染。实验结果表明该系统有良好的实时性和有效性。
Augmented reality is the false and true information fusion technology. With the popularity of smart phones,application development of the mobile platform augmented reality has become a hot spot. Because the mobile augmented reality system based on fiducial marker has some defects of mass calculation and poor rectification tracking realtime performance of identification detection recognition algorithm in the case of limit hardware calculation resource of smart phone,a scheme to detect four corner points quickly by means of the floodfill method and geometrical characteristics of bounding rectangle is proposed to reduce computation time. The fiducial marker based on hamming code was designed to improve the marker identification speed. In order to quickly and accurately compute the camera external parameters,the plane homography matrix was optimized by using LevenbergMarquardt iterative algorithm,and then the rendering is loaded on the virtual object on fiducial marker according to the obtained camera external parameters. The experimental results show that the system has perfect realtime performance and effectiveness.