Research And Application Of Transfer Function Specification In Direct Volume Rendering

Volume rendering is a key technology for the visualization of three-dimensional dataset. Transfer function is used to specify the relationships between volume data and optical properties in the process of volume rendering. Specification of transfer function plays an important role in the rendering quality. Unfortunately, specifying a good transfer function is a difficult and tedious task, thus it has been referred as one of the top ten problems in volume visualization.Recently experts have been managed to introduce intelligent algorithms to specification of transfer function. They consider the design process of transfer function as a parameter optimization problem, and realize intelligent specification of transfer function. However there are still existing some problems such as poor rendering results evaluation methods, exploration blindness for the large volume datasets and inefficiency in the intelligent specification. This paper studies the intelligent specification of transfer function, especially on the image evaluation module and the transfer function updating module. The content of our paper is as follows:As for the problem existing in rendering image evaluation module, e.g. evaluation standards of image quality are different and used unpopular, we present an image evaluation method of image information entropy. Based on the comparison on types of methods of image quality, Image information entropy was introduced into rendering evaluation module, utilizing its intrinsic characteristic of measuring the complexity of the image texture. It improves visualization effect of intelligent transfer function implementing on different volume data.We also study and compare the Genetic Algorithm(GA), Particle Swarm Optimization(PSO) and Quantum-behaved Particle Swarm Optimization(QPSO) algorithms, further verify the characteristic of QPSO, e.g. less parameter, fast convergence and excellent global optimality. Aiming to solve the design inefficiency in intelligent specification of transfer function, we present an updating method for transfer function based on QPSO, utilizing the QPSO to update the iteration process. The result achieves the real-time requirement, which shows excellent global optimal abilities of QPSO. Based on the above mentioned, this paper presents an intelligent specification method for transfer function based on QPSO, which refers image information entropy as the image quality evaluation method. Our method designs kinds of transfer functions on different volume datasets. We implement the method in a large-scale seismic data visualization system. We also provide users with both manual and automatic transfer function design tools, which is based on the intelligent algorithms GA and QPSO. The experimental results show that our method can get the satisfactory image that the geophysicists'users desired in exploration of seismic data, thus improve the accuracy of seismic and success rate.