Generally, image-based evaluation on the performance of textile appearance such as smoothness and pilling was interfered by fabric pattern, color, and illumination. In this paper, a new method was used to reconstruct a 3D surface of fabric based on image's features which could eliminate the influence of surface texture. Concretely, two parallel-placed digital cameras were calibrated based on planar pattern and the captured binocular images were rectified through epipolar line. The feature points

双目密集立体 使用双目立体的多视图重建

该程序利用大恒水星相机，两个相机同步启动，然后可以点击按钮同时保存图片，可以自己采集双目照片用于测试。

小秋SLAM笔记

VS2017+OpenCV通过双目视觉原理，通过SIFT特征点匹配恢复物体深度信息，输出三维坐标。