Edge artifact correction for industrial computed tomography images

To eliminate the edge artifacts of industrial CT images, and improve the identification ability of the image and the precision of the dimension measurement, a coefficient adjusting method for reducing crosstalk noise is proposed. It is concluded from theoretical analysis that crosstalk generated from adjacent detectors by Compton scattering is the major reason for the edge artifacts. According to the mathematic model of the detector crosstalk, we design a special detector system configuration and stair-step phantom for estimating the quantity of crosstalk noise. The relationship between crosstalk ratio and intensity of the incident X-ray is acquired by regressing experimental data with least square method. The experimental result shows that the first-order crosstalk ratio between detectors is about 9.0%, and the second-order crosstalk ratio is about 1.2%. Thus the first-order crosstalk is the main factor causing edge artifacts. The proposed method can reduce the edge artifacts significantly, and meanwhile maintain the detail and edge of CT images.