A Research To Sampling Method On Volume Rendering

Volume rendering is an important part in scientific visualization, and also a popular research field in computer science. In order to use volume rendering technique in practical, obtain high quality volume rendering results, we did some research and optimize in the sampling method of scalar value and gradient of the volume data.We summarized previous sampling methods, mainly on the scalar value interpolation and filtering. We did a brief introduction of the tri-linear interpolation, Gaussian filtering, moving least square and tri-cubic B spline. We also introduced some algorithms presented in recent years, such as virtual sampling, peak finding and second order pre-interpolation. By using these algorithms to reconstruct the scalar value, we can obtain scalar value faster, while avoiding the interpolation computing in voxel, thus improve the rendering efficiency. After all, these algorithms can obtain better rendering quality and rendering speed.Inspired by the adaptive sampling method and pre-integration algorithm, we presented a method called Adaptive Pre-Integration with Multiple Pre-Integration Tables. The key idea of this algorithm is that adaptive sampling and proportional decrease the step size by the scalar of volume data and the position along the ray traversal. After we got the information of divide of the sampling segment, we accumulate the color the opacity by different pre-integration with different step size. Experimental results demonstrate that our method can get better rendering results than uniformed step size of two-dimensional pre-integration table method; and better interaction than three-dimensional pre-integration table method.We also developed a volume rendering system with Qt, OpenGL and GLSL. Based on this in-house system, we did some further research of the algorithm, compared the difference of the rendering images. Also with this system, we realized many algorithms of volume rendering.