【摘要】 全景图是虚拟现实和计算机视觉中一种重要的场景表示方法 .通常获得高质量的全景图需要使用昂贵的专用设备 ,而且拍摄时需要精确地校准摄像机 .从普通摄像机图像拼接是获得全景图的一种低成本而且比较灵活的方法 .文中提出一种新的摄像机图像拼接算法 ,利用摄像机绕垂直轴旋转 36 0°依次拍摄的照片序列 ,拼接圆柱面全景图 .该算法不要求校准摄像机 ,对相邻帧之间摄像机的转角也没有严格的限制 ,而且不受帧间光照强度剧烈变化的影响 .从实验结果看 ,该算法获得了理想的拼接效果更多还原

【Abstract】 As an image representation of virtual environment, panorama has important applications in Virtual Reality and Computer Vision. Generally, to get a panorama with high quality, we need some special, expensive and carefully calibrated equipment. Image Mosaic is an inexpensive and flexible approach to get panorama. In this paper, a novel panorama composition algorithm is presented. By rotating a camera around a vertical axis and composing the video frames, our algorithm can create cylindrical panoramic images with ideal quality. Our algorithm needs no calibration, imposes no strict limit on camera rotation angle between consecutive frames. And at the same time, this algorithm is not sensitive to the exposure difference between consecutive frames.更多还原