There are many kinds of defects in pipelines, which are difficult to detect and of low automatic test capability. aiming at this issue, a new omnidirectional vision inspection method for the morphology defect detection in pipelines is presented in this paper. Firstly, an Omnidirectional Vision Sensor (ODVS) is designed to obtain the omnidirectional texture information of the pipeline interior. Then, an Active Stereo Omnidirectional Vision Sensor (ASODVS) is designed to obtain the depth information of the inner wall of the pipeline quickly with high precision. Finally, by combining these two kinds of ODVS and merging the omnidirectional measured data with the omnidirectional texture data by the software, the functional and structural defects of pipelines can be automatically analyzed and evaluated. Meanwhile the 3D reconstruction of the pipeline can be achieved. The experimental results show that the ASODVS+ODVS designed in this paper can obtain images with texture information and geometry information inside the pipeline in real time. Tthe pipeline robot can be configured to inspect and distinguish the defects, simultaneously reconstruct the pipeline with the robot walking.