Omnidirectional Imaging Sensor Based on Conical Mirror for Pipelines

As a sensor carrier and directional constraint equipment, precision pipelines are prone to significant damage to the inner surface over time, posing safety hazards. To implement the inspection of the pipeline's inner surface, an omnidirectional imaging sensor based on a conical mirror and single camera was designed in this paper. The optical imaging model of the sensor was constructed to analyze the field-of-view (FOV) and depth-of-field (DOF) of the sensor, and the parameters of the sensor's structure were determined. Then, a concentric circle unwrapping method based on secondary correction was proposed to correct the distortion of panoramic images. Additionally, the position of the conical mirror was calibrated to rectify the assembly error. Finally, unwrapped images and YOLOv8 were utilized to implement the detection of the defects, such as scratches and pits, on the inner surface of pipelines. The experiments showed that the accuracy of defect detection was approximately 80.1%. The omnidirectional imaging sensor designed in this paper was capable of imaging the inner surface of pipelines, which was used for detecting and classifying the surface defects.