Abstract: In direct volume rendering, it is very difficult to remove or suppress less important parts of a scene to reveal Region Of Interesting （ROI）. In order to enhance the region of interest, a context-preserving volume rendering is presented, using the spatial relationships between the sampling points and the focus （center points of ROI）. A focus is located manually to determine ROI. Then the spatial information including the distance between sampling points and focus, the angular separation is gathered and fed to a pre-defined Gaussian-attenuation function. Finally, this attenuation function is used to control the composition of the rays. The proposed method is evaluated on CT head data. The experimental results show that occluding objects are rendered sparsely or suppressed entirely, at the same time, the ROI is visually emphasized and preserves the clear shape cues of background and good human-computer interaction.

Key words: direct volume rendering, context-preserving, ray-casting algorithm, focus, transfer function, Gaussian function

摘要： 直接体绘制中，为了去除非感兴趣组织对感兴趣区的遮挡，增强感兴趣区域的显示效果，提出利用空间信息的保留上下文体绘制方法。通过人机交互设定焦点，快速定位感兴趣区;提取光线上每个采样点与焦点的空间位置以及夹角信息;根据这些信息构建高斯衰减函数，作用于当前采样点的不透明度，控制光线合成，以突出显示感兴趣区。对CT头部数据的实验结果表明，该方法可以去除非感兴趣组织对感兴趣组织的遮挡，突出显示感兴趣区，且具有良好的人机交互性。

关键词: 直接体绘制, 保留上下文, 光线投射算法, 焦点, 转换函数, 高斯函数