Research On Calibration Method And Limited-angle Reconstruction Algorithm For CT System

Computed tomography technique has made it possible to detect the interior structure of object without any damage to the scanned object. With the rapid development of medical and industrial detection, the amount of CT being used is becoming bigger and bigger and the requirement for precision of CT is getting higher and higher. The misaligned geometry between the actual CT imaging system and the ideal mathematical reconstruction model, including X-ray source deviation, detector deviation and rotation axis deviation, can cause calculation errors of projection address in CT reconstruction, then reduce the imaging quality of reconstructed images severely. Therefore, calibration of CT imaging system in advance is an important and necessary step.Based on the pinhole camera model, a geometric calibration method for cone-beam and helical cone-beam CT imaging system is proposed and a calibration phantom with symmetric repeat texture is designed in this paper. Firstly, angular deviation of the detector can be made equivalent to rotation matrix of system parameters. Then based on the corresponding relation of each pixel between phantom projection in misaligned detector and aligned detector, calculating detector angular deviation can be turned into a minimum problem of the rank of calibration phantom image, which subjects to some constraints. Then the geometrical position of CT imaging system can be calculated based on the geometric relationship and imaging model of calibration phantom, which includes the distance between X-ray source and rotation axis, the distance between X-ray source and detector, the projection coordinate of central ray and angle deviation between rotation axis and lift axis. The proposed method uses global features of projection image of the calibration phantom and has higher accuracy. Finally, this paper carries on practical experiments.As CT devices are updating quickly and a lot of improved reconstruction algorithms have been proposed, CT technology has achieved rapid development. In practical CT detection applications, detected objects in industrial CT are diverse. In consideration of the X-ray radiation energy levels, the geometry structure of the target, detector size and the detection space, we can just only obtain projection data in limited-angle. Thus, the research on limited-angle CT reconstruction is particularly important. The proposal of Compressed Sensing in 2006 indicates that one original signal can obtain ideal results by the lower sampling data than Nyquist sampling rate. Once this theory is proposed, a lot of researchers have been working on the combination between Compressed Sensing and limited-angle CT reconstruction. Based on the former research on limited-angle CT reconstruction algorithm, a limited-angle CT reconstruction algorithm is proposed on weighted anisotropic TV and wavelet regularization. The algorithm fully considers the direction of edge of image information and the sparse property of detected objects in wavelet domain. Simulation and real-data experiments under different conditions are implemented to verify the effectiveness and robustness of the proposed method.