PercGestalt2005

Visual Pattern Perception: 

Visual Pattern Perception

Content: 

Content Human Visual System constructs Reality Gestalt Approach (Psychology) Marr Approach (Computer Vision) Current Research Topic Your chance to become famous and rich!

What is that?: 

What is that?

Hups!: 

Hups!

Visual Illusion:Kanizsa Triangle: 

Visual Illusion: Kanizsa Triangle

Visual Illusion:Kanizsa Squares: 

Visual Illusion: Kanizsa Squares

Other Illusionary Contours:: 

Other Illusionary Contours:

Illusionary Spiral: 

Illusionary Spiral

Switching between ConceptsNecker Cube: 

Switching between Concepts Necker Cube How could we make the Necker Cube unique?

Necker Cube: 

Necker Cube Occlusion

Necker Cube: 

Necker Cube Perspective?

Motion Binding: 

Motion Binding

What is that?: 

What is that?

Marroquin: Visual System constructs permanently new Organisations: 

Marroquin: Visual System constructs permanently new Organisations

Summary: 

Summary Visual System creates Percepts Sometimes these Percepts are missleading -> Illusions Question I: Why does the Visual System do that? Question II: What are the underlying Mechanisms? Question III: How can they be modelled?

Question I: Why does the Visual System do that?: 

Question I: Why does the Visual System do that? The Visual System is badly designed?

Question I: Why does the Visual System do that?: 

Question I: Why does the Visual System do that? The Visual System is badly designed? Well, there is no better system on the market right now!

Question I: Why does the Visual System do that?: 

Question I: Why does the Visual System do that? The Visual System is badly designed? Well, there is no better system on the market right now! The Visual System makes use of likely to be expected events that reflect Interdependencies in visual Data!

Question I: Why does the Visual System do that?: 

Question I: Why does the Visual System do that? The Visual System is badly designed? Well, there is no better system on the market right now! The Visual System makes use of likely to be expected events that reflect Interdependencies in visual Data! Why does it do that?

Question I: Why does the Visual System do that?: 

Question I: Why does the Visual System do that? The Visual System is badly designed? Well, there is no better system on the market right now! The Visual System makes use of likely to be expected events that reflect Interdependencies in visual Data! Why does it do that? Because visual Data is ambiguous!

Question II: What are the underlying Mechanisms?: 

Question II: What are the underlying Mechanisms? Gestalt Psychology Max Wertheimer Untersuchungen zur Lehre der Gestalt (1923) Kurt Koffka Principles of Gestalt Psychology (1935) Wolfgang Köhler Gestalt Psychology: An Introduction to new Concepts in modern Psychology (1947) The whole is more than the sum of its parts Laws about Perceptual Organisation

What is the underlying Reason for Feature Binding?: 

What is the underlying Reason for Feature Binding?

Law of Common Fate: 

Law of Common Fate Visual Entities that move in a common Direction belong to the same Object

What is the underlying Reason for Grouping?: 

What is the underlying Reason for Grouping?

Slide26: 

Law of Good Continuation Visual Entities that are organised along a smooth Line or Curve belong to the same Object

What is that?: 

What is that?

What is that?: 

What is that?

What is that?: 

What is that?

What is the underlying Reason for Grouping?: 

What is the underlying Reason for Grouping?

Law of Proximity: 

Law of Proximity Visual Entities that are close to each other belong to the same Object

However!: 

However!

What is that? : 

What is that?

Slide34: 

Law of Proximity Visual Entities that are close to each other belong to the same Object Law of Proximity Visual Entities that are close to each other belong to the same Object Law of Proximity Visual Entities that are close to each other belong to the same Object Law of Similarity Visual Entities that are simlar belong to the same Object

What is that?: 

What is that?

Law of Symmetry: 

Law of Symmetry Visual Entities that are symmetric are tend to be seen as joint entities

Slide37: 

What is that?

Slide38: 

Law of Closure Visual Entities are tend to be seen as closed figures

Slide39: 

Law of Prägnanz (Koffka) Of several geometrically possible organisations the one will actually occur which possesses the best, simplest and most stable shape

Slide40: 

Law of Prägnanz (Koffka) Of several geometrically possible organisations the one will actually occur which possesses the best, simplest and most stable shape What is the problem of this law?

Slide41: 

Law of Prägnanz (Koffka) Of several geometrically possible organisations the one will actually occur which possesses the best, simplest and most stable shape What is the problem of this law? Is there a common thing in all these Gestalt laws?

Law of Prägnanz: 

Law of Prägnanz

Law of Prägnanz: 

Law of Prägnanz

Hups: 

Hups

Is this Illusion explainable by Gestalt Laws?: 

Is this Illusion explainable by Gestalt Laws?

Is this Illusion explainable by Gestalt Laws?: 

Is this Illusion explainable by Gestalt Laws? Importance of Object Knowlegde!

Problem in the Formulation of Gestalt Laws in the book!: 

Problem in the Formulation of Gestalt Laws in the book! Example: Law of Similarity Book: The law of similarity states that objects similar to each other tend to be seen as a unit Lecture: Visual Entities that are simlar belong to the same Object

Marr's Computational Theory of Vision: 

Marr's Computational Theory of Vision David Marr was very influentual to modern Approaches to model Vision Marr 1982. Vision: A computational investigation into the human representation and processing of visual information 3 Stages Primal Sketch: Multi-scale Edge Detection 2.5D Sketch: Viewer centered Scene Representation 3D Sketch: Object Centered Representation

Primal Sketch: 

Primal Sketch

2.5D Sketch: 

2.5D Sketch

3D Sketch: 

3D Sketch

Current State of the Art: 

Current State of the Art People are trying hard more than 1000 groups world-wide Denmark (appr. 10 groups, 4 in Copenhagen) Germany (more than 80 groups) United Kingdom (more than 100 groups) more than 10.000 researchers world-wide large interdisciplinary effort Psychology, Neurophysiology, Mathematics, Computer Science, Computational Neuroscience, Engineering, Visual Arts there is a lot of money involved

Current State of the Art: 

Current State of the Art Artificial Visual Systems become established in controlled domains e.g., industrial environments with controlled illumination specific applications using large amount of prior knowledge person tracking with known-background pose-estimation of known objects However, there are no artificial visual systems that are able to solve similar tasks than humans

So, it is your chance!: 

So, it is your chance!

Exercises: 

Exercises

Exercise I: 

Exercise I Create a Drawing in which the perceptual Organisation according to specific Gestalt laws becomes visible Law of Similarity Law of Proximity Law of Prägnanz

Exercise II: 

Exercise II What Gestalt Laws are involved in the Kanizsa Illusion

Exercise III: 

Exercise III Create your own Kanisza Illusion

Exercise V: 

Exercise V Give a precise Definition of the Gestalt Law of Pregnanz Of several geometrically possible organisations the one will actually occur which possesses the best, simplest and most stable shape What is the meaning of best simplest stable

Exercise IV: 

Exercise IV Create a visual Illusion that makes use of Law of Similarity Law of Proximity Law of Good Continuation