Research And Application On Visual Computational Theory In Frequency Domain

Visual Computational Theory which was founded by D.Marr has made great contribution on machine vision and AI(artificial intelligence), and it also made great influences on our research of human visual system. D.Marr has done lots of work for Visual Computational Theory, but in spatial frequency domain, it is not enough. In fact, it is certain that human visual system has special spatial frequency characteristic, so combining Visual Computational Theory with Spatial Frequency is the essential way to perfect the theory.We made research in Visual Computational Theory, and combined it with Receptive Field, Visual Multi-frequency Channel Theory and visual contrast sensitivity function to bring the concept of spatial frequency into the theory. Our main purpose is to build Visual Computational Theory in spatial frequency domain and perfect the theory. And the main work and some highlights of this thesis are summarized as follows:1. Researching on receptive field, Visual Computational Theory and Visual Multi-frequency Channel Theory, to build the math model of Marr operator in spatial frequency domain. Based on Physiology result, we will use Marr operator in spatial frequency domain to simulate the Contrast Sensitivity Function.2. Using Marr operator in spatial frequency domain to resolve the visual problem that Marr operator can not resolve. For example, some visual illusions detetion. Visual illusions come out from our visual system but can hardly be detected by computers, which is why it is important to study these illusions to know more about our visual system. And it is a certain way to prove the great visual computational ability the new operator has.3. Using Marr operator in spatial frequency domain to reinterpret the Visual Integrity which is not well interpreted in Visual Computational Theory. In order to reinterpret the Visaul Integrity, we reinterpret Chen’s famous visual experiment by Visual Computational Theory in spatial frequency domain. In this way, we can reinterpret the Visual Integrity.After all, we have only one visual system, and there’s only one principle in it. After combining Visual Computational Theory with spatial frequency and lots of experiment done, we can successfully prove the great visual computational ability the new theory has, and we can also reinterpret the Visual Intigrity which Visual Computational Theory hardly can do. In this way, some contribution can be done for perfecting Visual Computational Theory.