Rapid prejudgment of reconstructed object volume and its adaptive reconstruction for industrial cone-beam CT

For the problem of mass redundant computation usually existed in the traditional reconstruction of industrial cone-beam CT, a rapid prejudgment method of reconstructed object volume is proposed. According to the analysis of the principle of Z-line data priority reconstruction, it is feasible to conduct a selectively sectional reconstruction. So a corresponding method is proposed to prejudge the reconstructed object volume rapidly, and relative parameters are calculated by considering both the definition of the rectangular bounding box and the projection relationship in cone-beam CT scan. Then the computational redundancy is reduced greatly. For its implementation in cone-beam CT, the parallel computing capability of the multi-core CPUs is fully performed by combination of two strategies, adaptive CPU cores allocation and reconstructed volume partition, based on the estimation of computation task. The experiment results show that the proposed method has good adaptability and significantly higher speed compared with Z-line data priority reconstruction. The proposed method can also find its application in the CT systems which employ the analogous reconstruction algorithms.