A Perceptual Framework for Comparisons of Direct Volume Rendered Images

Abstract. Direct volume rendering (DVR) has been widely used by physicians, scientists, and engineers in many applications. There are various DVR algorithms and the images generated by these algorithms are somewhat different. Because these direct volume rendered images will be perceived by human beings, it is important to evaluate their quality based on human perception. One of the key perceptual factors is that whether and how the visible differences between two images will be observed by users. In this paper we propose a perceptual framework, which is based on the Visible Differences Predictor (VDP), for comparing the direct volume rendered images generated with different algorithms or the same algorithm with different specifications such as shading method, gradient estimation scheme, and sampling rate. Our framework consists of a volume rendering engine and a VDP component. The experimental results on some real volume data show that the visible differences between two direct volume rendered images can be measured quantitatively with our framework. Our method can help users choose suitable DVR algorithms and specifications for their applications from a perceptual perspective and steer the visualization process.