Automatic Reconstruction of Sewer Shaft Profiles from Fisheye-Lens Camera Images
The focus of this research project in cooperation with IBAK Helmut Hunger GmbH & Co. KG is the automatic shape measurement of sewer shaft profiles from image sequences acquired by a fisheye-lens camera hanging into the shaft.
The camera is provided with an inertial sensor which measures rotation around the camera's viewing axis. Images and sensor data are used to estimate the pose of the camera inside the shaft and measure 3D points on the surface of the shaft. Afterwards, 2D shape fitting techniques are used to estimate the shape of the shaft profile at different depth. From profile contours the 3D geometry of the shaft is computed which can e.g. be used to generate a wire-frame model of the shaft. By applying a problem-specific geometric correction to the reconstruction, we are capable to reconstruct even very long sequences (>10 m) robustly even in case of heavy illumination changes and sudden geometry changes of the shaft.
An implementation of our developed approach has been integrated into the software for the commercial sewer shaft inspection system IBAK Panoramo® SI . It has been applied successfully to the practical stage and has delivered useful results, as e.g. presented by the Sanitation Department of Göttingen in .
A research paper about this project has been accepted to the 31st annual symposium of the German Association for Pattern Recognition (DAGM 2009) and has been published in the proceedings .
The project is funded by IBAK Helmut Hunger GmbH & Co. KG, Kiel.
 S. Esquivel, R. Koch, H. Rehse: Reconstruction of Sewer Shaft Pro?les from Fisheye-Lens Camera Images, in: Lecture Notes in Computer Science 5748 (DAGM 2009), pp. 332–341, 2009 (available at www.springerlink.com)