Perceptual image rendering

• Main Author: Low, Yih Fei.
• Other Authors: Lin Weisi
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Computer graphics
• Online Access: http://hdl.handle.net/10356/49135
• Institution: Nanyang Technological University

Given a limited computation budget, perceptual image rendering aims to produce synthesized images with higher perceived quality than uniformly rendered ones by exploiting the properties of the human visual system (HVS). There are several perceptual models that have been used in the rendering process, such as the human visual attention (VA) model and the human visual system (HVS) model. In this project, the just noticeable difference (JND) model is explored. The JND model represents the change of the sensitivity of the HVS with the spatial frequency of image contents. The sensitivity decreases on high frequency patterns and this allows a higher level of error to be tolerated. Different from the VA model which describes a global characteristic of the image contents, the JND model describes a local characteristic of the image contents. Therefore, a perceptual model is proposed by combining the JND model with the VA model and the combined VA-JND model is applied in the adaptive sampling process of image rendering. Experiments have been performed to analyse the difference between images rendered perceptually with the VA model only and images rendered with the combined VA-JND model. The experimental results that the combined VA-JND model can further improve the quality of images rendered with the VA model only, within a given sample budget.