Efficient Image Rendering Using a Fuzzy Logic Model of Visual Attention
Ross Andrew Brown, Binh Pham, Eugene Aidman, Anthony John Maeder · QUT ePrints (Queensland University of Technology) · 2000
Physiological research has revealed constructs named “feature detectors‿ in the human visual system. These detectors react to differences in visual features, marking them as salient and thereby attracting the attention of the viewer. In this paper we outline a fuzzy logic system which processes 3D scene descriptions to compute the relative visual importance of regions in the scene. We detail experiments carried out to obtain parameters for the model, and the methodologies used in the aggregation, implication and defuzzification processes. We finish with a discussion of future work for further refinement of the model. The system is expected to have applications in the area of efficient image synthesis, in particular, progressive rendering and transmission techniques.