Abstract

Info

Title:: Ray casting algorithm for three-dimensional reconstruction based on improved pre-integration technology

Author(s):: WANG Yachen; WANG Yu; GONG Ning; XIAO Hongbing; School of Artificial Intelligence, Beijing Technology and Business University, Beijing 100048, China

Keywords:: image reconstruction ray casting algorithm pre-integration technology

Abstract:: Due to the nonlinearity of the transfer function in the traditional ray casting algorithm, Nyquist sampling theorem often cant be satisfied, which affects the quality of the generated model. Therefore, a novel ray casting algorithm is proposed for three-dimensional (3D) reconstruction based on improved pre-integration technology. The pre-integration classification technology is introduced into the traditional ray casting algorithm to improve the quality of the reconstructed 3D model by avoiding the maximum Nyquist sampling frequency for any nonlinear transfer function. Then, the pre-integration acceleration operator is used to optimize the calculation methods for opaqueness and color value, thus simplifying the pre-integration synthesis process and shortening the time for 3D model reconstruction. The proposed algorithm speeds up the reconstructions for the synthetic 3D cone and real coronary artery CT image by 22.5% and 23.8%, and additionally decreases mean square error of the generated 3D cone images, while increasing structural similarity index as compared with the traditional ray casting algorithm, which demonstrates that the proposed algorithm can improve both the reconstructed model quality and the reconstruction speed.