Investigation On The Design Of Transfer Functions Based On Ant Colony Algorithm

Currently, the CT technology has been developed into a relative mature stage, however, the technology for the DICOM data visualization techniques which is far behind the development of the CT technology due to the huge and complex data has become a barrier for the application of the CT technology. The computing speed of the visualization techniques is improved a lot on the basis of the GPU operation as the computerized hardware technology rapidly developed in the recent decade years, thereby the main position of the speed issue is replaced by the transfer function which is an important segment of the visualization techniques.In this paper, the intelligent design based on the intelligent optimization algorithms is realized, improving the efficiency of the volume rendering algorithms, and avoiding the trouble of the transfer function in the process of the manual and semi-automatic operation. meanwhile reducing the requirement for the level of the professional staff. Moreover, automatic transfer function design could render data more concisely and objectively.The main content in this paper is as follow:(1) The improvement on the ant clustering analysis is made on the basis of the transfer function design of the image. And we propose the method for the transfer function design based on the ant clustering analysis. This method which is extended from two-dimensional makes the transfer function design automatically. It cuts down the difficulty level for the users, reducing the iterations of the transfer function design, improving the design efficiency.(2) Furthermore, as the powerful ability of GPU at the respect of the general compute and graphic process is used to rapidly realize the volume rendering technology on the basis of the ant clustering analysis, optical model and graphics pipeline is finished with GPU hardware, therefore the speed of volume rendering is improved a lot.(3) The space of data occupied memory is reduced due to the use of the octree method which can search and locate data. This structure of data makes space skipping possible and reduces much memory, therefore the limit of edge-connectivity based on GPU depth buffer is overcome.In this paper, the modified ant clustering analysis algorithm is realized by coding the program and applied to the three dimension medical data classification, leading to the realization of the transfer function intelligent design. Meanwhile, the visualization efficiency is improved remarkably due to the acceleration of the hardware and the application of octree, providing the instructive exploration for the ant colony algorithm at the respect of the transfer function design.