Abstract:Textons, coined by Julesz in the 1960s, refer to the atomic structures in natural images and are considered the basic elements in early (pre-attentive) visual perception. The primal sketches, conjectured by Marr in the 1970s, refer to parsimonious token representation of generic images. Both concepts are fundamental to computer and human vision studies. However they have been very elusive due to the lack of rigorous mathematical definitions and models. The objective of this dissertation is to study a mathematical framework for the two concepts and thus to build a theoretical foundation for the early vision theory. Our study leads to generative image models for generic natural images and computing algorithms for learning, simulation, and inference. The framework integrates two main statistical modeling paradigms in the literature. (i) Hierarchic generative methods, such as transformed component analysis, wavelet coding, and sparse coding; and (ii) Descriptive methods, such as Markov random fields, graphical models, and minimax entropy learning. Generally speaking, the former decompose images into image elements, and the latter represent spatial relationships between the components. The dissertation makes the following contributions. (1) We define textons and image primitives as elements in the dictionaries of generative models which are learned from natural images through maximum likelihood estimation. (2) We study a Gestalt field for modeling the spatial relations among textons with reconfigurable neighborhoods. (3) We present a primal sketch model by dividing an image lattice into structural parts as textural parts which are modeled by the generative models and descriptive models respectively. (4) We study the information scaling phenomena and the transitions between various regimes of image representations, for example, the transition between textons and textures.

Outline of a Theory of Visual Tension

Outline of a Theory of Visual Tension

Tension in Einstein's Philosophical Thinking

The Equilibrium and Form-Finding of General Tensegrity Systems with Rigid Bodies

Self-Equilibrium, Mechanism Stiffness, and Self-Stress Design of General Tensegrity with Rigid Bodies or Supports: A Unified Analysis Approach

A Unifying Framework for Form-Finding and Topology-Finding of Tensegrity Structures

Tension in Modern Art in Perspective of Paradigmatic Turn

Outline of a Theory of Scientific Aesthetics

Theory of Artistic Vision:A Chinese Version of German Aesthetics

Discussion on the Tension Application and Performance of Line in Design

Arnheim's Gestalt Theory of Visual Balance: Examining the Compositional Structure of Art Photographs and Abstract Images

A Case Study on Formative Elements and Visual Perception Theory Expressed in Virtual Art Contents

Development of the Visual Cognitive Theory and Its Application in Architectural Design

Aesthetic Assessment of Paintings Based on Visual Balance

Art's Hidden Topology: A window into human perception

An eye tracking study of the application of gestalt theory in photography

A framework connecting vision and depiction

A Mathematical Theory of Textons and Primal Sketch: Integrating Generative and Descriptive Methods

Aesthetics of tension

Research on Visual Attraction Elements and Mechanism of Landscape Space