On a new calibration method for single viewpoint constraint for catadioptric omnidirectional vision

The single viewpoint constraint (SVC) is the principal optical characteristic for most catadioptric omnidirectional vision (COV). Precise calibration for SVC is needed during system assembling. Owning to the nonlinear distortion in the imaging system, the precision of calibration based on ideal perspective imaging model is often poor. A new calibration method of SVP for the COV was proposed. Firstly, an image correction algorithm was obtained by training a neural network. Then, according to characteristics of space circular perspective projection, the corrected image of the mirror boundary was used to estimate whose position and attitude relative to the camera to guidance calibration. Since the estimation was conducted based on actual image model rather than the simplified model, the estimation error is largely reduced, and the calibration accuracy is clearly improved. Experiments are conducted on simulated images and real images to evaluate the accuracy and the effectively of the method.