Abstract

Visual perception is the ability to interpret, process, and comprehend all the information received through the sense of sight by association with earlier experiences. Researchers have long struggled to explain what visual processing does to create what we actually see, and brought many theoretical approaches explaining how human beings see the world. The theoretical approaches of visual perception differ widely and their coverage ranges from early theories such as Gestalt theory to recent computational theory in the field of Artificial Intelligence. According to the characteristics of visual perception, human beings tend to classify the ambient environment objects into different categories described by various symbols or objects. Similar symbols or even quite dissimilar symbols may be perceived as belonging together or belonging to different groups according to people's judgment. It must follow certain rules when human beings set up relationships between those objects and symbols, and finally obtain the unambiguous perceptual results through the process of visual perception. To find out the mechanisms underlying these properties of visual perception, this present thesis conducts experiments on perception using curve reconstructions as test cases. The perception model developed through the experiment is implemented in a curve reconstruction algorithm. It is assumed that a good perception model will reconstruct curves in the same manner as human beings perceive them. In the present thesis, a series of methods from Design of Experiments (DOE), ANOVA and the multivariate nonlinear regression model are applied to investigate the relationships between the points and curves. The results show that our perception model conforms to the pattern human perceives the points.