Accelerated Techniques Study On Real-time Ray Casting

Scientific visualization has been one of the most rapid developed techniques used in the applications of medicine, meteorology, geology and so on in recent years. Volume rendering is a key technique of scientific visualization. It has advantages in observing structures of three-dimension datasets. However, volume rendering algorithms still suffer from high computational complexity and low rendering speed. So the accelerating techniques need to be further studied, including both hardware-based and software-based ones, which are the main topic of this paper.There is an increasing need for improving GPU-based volume rendering algorithms in recent years because GPU (graphics process unit) has many advantages in getting better rendering performance. Nowadays GPU is one of the most frequently used techniques for accelerating volume rendering algorithms. As GPU is also the basic element in the later experiment, the implementations and improvements of texture-based algorithms are firstly presented in this paper. A new sampling method is adopted to calculate texture coordinates of the samples by taking advantage of powerful matrix functions of the modern GPU. Additionally, dependent texture technique, which maps the scalar value of sample to optical color and opacity in the GPU, is also employed in order to save texture memory. The texture-based algorithms are divided into two categories: 2D-texture-based algorithm and 3D-texture-based algorithm. These two algorithms are respectively discussed and compared with each other. The experiments results show that the algorithm using new sampling method produces better rendering performance and image quality than traditional texture-based algorithms for 2563 dataset.The traditional ray casting algorithm which is the most accurate volume rendering algorithm suffers from high computational complexity. In order to accelerate the traditional algorithm, a new ray casting accelerating algorithm, which is called SRC (segment-based ray casting) for short, is proposed in this paper. There are many optimization techniques making use of data coherence to reduce process time in former researches. SRC works the same way, but in the different rendering stage. SRC combines sequential samples which have the similar optical values into a segment. The segments substitute for the samples to be the basic composition units. Therefore, the number of composition operations is reduced. The efficiency of SRC is...