This paper presents a box detection system for small box depalletizing. The system uses a structured-light color depth camera, ROS, PCL, OpenCV, and their Python wrapper modules. Repetitive RANSAC plane fittings and Euclidean clustering extract planar...
This paper presents a box detection system for small box depalletizing. The system uses a structured-light color depth camera, ROS, PCL, OpenCV, and their Python wrapper modules. Repetitive RANSAC plane fittings and Euclidean clustering extract planar surfaces from cloud, and minimum area rectangle estimations on both 3D and projected 2D concave hulls find box surfaces. After filtered by discontinuity checks in the gray depth image, lines in the projected color image successfully divide densely palletized box boundaries.