A New Technique for Camera Self-Calibration

Self camera calibration based on purely rotational movement of camera receives the most attention among different calibration methods. The purely rotational approach, however, assumes the camera rotates around its optical center, therefore yielding no translation. This assumption is not realistic since in practice the precise location of optical center is often unknown and rotation is often performed about an unknown but fixed point near the optical center. The rotation is therefore not pure in the sense that relative translation is introduced between two rotated camera frames. In this paper, we introduce a new rotation-based self camera calibration method, which explicitly accounts for the unknown translation. The problem is mathematically formulated and solved taking the translation into consideration. Experiments with both synthetic and real data show that the systemic errors from the ignorance of the translation can be effectively eliminated by our approach.

Authors

• Qiang Ji
• Songtao Dai

Keywords

• Cameras
• Calibration
• Optical sensors
• Robot vision systems
• Computer vision
• Layout
• Computer science
• Systems engineering and theory
• Computer errors
• Robot kinematics
• Camera Self-calibration
• Systematic Errors
• Fixed Point
• Camera Frame
• Principal Plane
• Camera Calibration
• Camera Rotation
• Pure Rotation
• Coordinates Of Points
• Image Frames
• Image Point
• 3D Point
• Visual Perspective
• Coordinate Frame
• Intrinsic Parameters
• Calibration Results
• Image Coordinates
• Object Point
• Translational Movement
• Rotation Point
• Camera Intrinsic Parameters
• Nelder-Mead Algorithm
• Calibration Algorithm
• Unknown Point
• Point In Frame
• Translational Impact
• Calibration Purposes
• Relative Orientation