For the engineering problem of low efficiency of defects inspection and assessment of pipe with existing method, an in-pipe internal inspection system based on active stereo omnidirectional vision sensor(AODVS) was presented to acquire 3D coordinates of point cloud and detect the defects on the inner surface of pipes in real time. First, inner surface panoramic images and laser streak panoramic images were captured with AODVS, Inner surface images were processed as follow: unwrapping, preprocessing, feature extracting and defects classification, Laser streak images reflecting the shape of inner pipe were processed to calculate 3D coordinates of the point cloud of inner surface. Finally, the pipe′s triangular grid model with real texture information was reconstructed by 3D modeling technique. Experiment results show the efficiency of proposed method to detect racial variation, holes, cracks and corrosions with high accuracy, this system provide a new online inspection approach to 3D measurement and reconstruction of industrial pipes.