一种空间自适应采样和递归线性估计的光线投射算法

A Spatial Adaptive Sampling and Iterative Linear Estimation-based Ray Casting Algorithm

光线投射算法因其成像质量高而广泛地用于虚拟内窥镜系统,但成像速度非常缓慢。为此,本文提出了一种自适应采样和递归估计的成像加速算法。首先,根据查找法快速得到的梯度和光线方向信息,自适应地调整采样步长,使得该算法能够以大步长快速跳过体素值变化缓慢的区域,同时在体素值变化剧烈或快接近等值面的区域,能够以小步长进行搜索。其次,以递归线性插值的方法估计投射光线与实际等值面的交点,用于补偿大步长导致交点精度的降低,此举能够显著地提高成像质量。实验结果表明,该算法在保证绘制图像质量的前提下,提高了体绘制速度,取得了比较满意的效果。

Ray casting is widely applied in virtual endoscope systems, due to high quality images rendered by the algorithm, but it is time-consuming. So we present a ray casting algorithm based on spatial adaptive sampling and iterative linear estimation. First, according to the gradient of volumetric data quickly obtained by a lookup table and ray direction, the algorithm automatically adjusts searching steps along the rays. The algorithm quickly traverses regions containing small variation of voxel values at large steps, and it searches at small steps when a ray encounters large varying areas. Second, to improve the efficiency of ray-isosurface intersections, an iterative linear estimation is proposed to obviously enhance rendered image quality. Experiments show the algorithm speeds up rendering speed and does not degrade image quality at all.