Pipeline Scene Reconstruction Based On Image Mosaicing

Robots should be able to perceive the surroundings in the complicated and unknown environment before carrying out further navigation. Consequently, environmental reconstruction is the premise for the robot autonomous operations. In this paper, a pipeline scene reconstruction method based on image mosaicing is proposed for cylindrical pipeline environment. With a wide-angle camera, the image sequence of the pipeline environment is captured. In order to obtain intuitional environmental information around the pipeline, an unwrapped model is proposed to unfold the distorted raw image to corrected flat surface image. By utilizing ORB (Oriented FAST and Rotated BRIEF) and weighted smoothing blending algorithm, image mosaicing with sequence frames are performed to realize scene reconstruction. The experimental results demonstrate that the proposed algorithm can achieve seamless stitching of pipeline image, and the number of keypoints is prominently decreased in comparison to that of FAST operator, while the quality of keypoints is improved. Compared with the classical SIFT and SURF operator, the time-consuming of the algorithm is improved about 2.5 times, which is more suitable for real-time environmental reconstruction.