基于对偶绝对二次曲面像变换的自定标方法

现有的自定标技术一般都是基于对偶绝对二次曲面像,由于对偶绝对二次曲面像元素在数量级上相差较大,在求解时必然会导致鲁棒性较差。提出了一种变换法,使对偶绝对二次曲面像的各个元素变到同一个数量级上,可以使求解过程更具有鲁棒性。模拟实验和真实实验结果表明,通过提出的变换方法能够极大地提高自定标的鲁棒性,而计算量并不会增加。

Robust serf-calibration method based on image of dual absolute quadric transformation

The available camera self-calibration is based on the image of the dual absolute quadric, but its elements have so large differences in magnitude that solutions are extremely sensitive to noise. To solve the problem, the method that all the elements of the image of dual absolute quadric are transformed into the same magnitude is proposed. So the solutions become more stable. The theorical analysis and experiments with both simulated and real data demonstrate that this self-calibration method can lead to an enormous improvement on the stability and robustness of the results without increasing computation.