Detection of morphology defects in pipeline based on 3D active stereo omnidirectional vision sensor

• Published in: IET image processing Vol. 12; no. 4; pp. 588 - 595
• Main Authors: Yang, Zhongyuan, Lu, Shaohui, Wu, Ting, Yuan, Gongping, Tang, Yiping
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 01.04.2018
• Subjects: 3D active stereo omnidirectional vision sensor; automatic optical inspection; camera motion estimation; detection network training; faster region proposal convolutional neural network; faster R‐CNN; fine detection; image reconstruction; image sensors; image texture; laser points space location information; learning (artificial intelligence); mechanical engineering computing; morphology defect detection; motion estimation; neural nets; omnidirectional vision inspection system; pipe 3D space reconstruction; pipe defect image database; pipeline inner wall depth information; pipeline inner wall texture information; pipelines; Research Article; stereo image processing; pipe 3D space reconstruction; camera motion estimation; omnidirectional vision inspection system; image reconstruction; pipeline inner wall texture information; laser points space location information; image texture; automatic optical inspection; image sensors; faster R-CNN; pipelines; mechanical engineering computing; stereo image processing; motion estimation; detection network training; faster region proposal convolutional neural network; pipeline inner wall depth information; learning (artificial intelligence); 3D active stereo omnidirectional vision sensor; morphology defect detection; neural nets; fine detection; pipe defect image database
• ISSN: 1751-9659; 1751-9667
• DOI: 10.1049/iet-ipr.2017.0616

There are many kinds of defects in pipes, which are difficult to detect with a low degree of automation. In this work, a novel omnidirectional vision inspection system for detection of the morphology defects is presented. An active stereo omnidirectional vision sensor is designed to obtain the texture and depth information of the inner wall of the pipeline in real time. The camera motion is estimated and the space location information of the laser points are calculated accordingly. Then, the faster region proposal convolutional neural network (Faster R-CNN) is applied to train a detection network on their image database of pipe defects. Experimental results demonstrate that system can measure and reconstruct the 3D space of pipe with high quality and the retrained Faster R-CNN achieves fine detection results in terms of both speed and accuracy.