An interval subdividing based method for geometric calibration of cone-beam CT

This paper presents a new method based on interval subdividing for geometric calibration in cone-beam CT system. The proposed method uses experimental data and requires no prior knowledge of the object. The geometric system is parameterized at first and the two critical misalignment parameters, that is, the transversal shift and the skew of the detector, are varied until achieving their values with high accuracy, which depends on maximizing the value of Q proposed in the method. To validate this method, numerical phantoms with and without noise are simulated respectively. The results demonstrate that this method is capable of achieving high accuracy for the parameters and is robust to the noise. Each parameter can be aligned utilizing this method individually. And this method has advantages in high resolution CT, in which many factors can affect the stability of the mechanical system. The results of reconstructions for a machine part demonstrate that this method can calibrate the parameters with high accuracy and converge quickly.