Class Lecture Notes on Motion and Depth Processing
Class Lecture Notes on Motion and Depth Processing
Motion Processing Overview
Introduction to Motion
Beginning discussion at 2 PM, hints of humor about being uncool.
Objective: Finish motion concepts and transition to depth cubes.
Global motion and depth processing are key topics.
Global Motion
Recap of Last Session
Discussed Reichardt detectors, which see motion locally but lead to the aperture problem.
Aperture problem: true motion direction is obscured due to limited view of edges.
Example: GIF showing ambiguous direction perception.
Reichardt Detectors
They are excellent at detecting local motion but have limitations relating to overall direction determination.
Motion detectors in area V1 receive visual input but need to combine local inputs for a global motion understanding.
Global Motion Detection
Combining information from multiple Reichardt detectors helps solve the aperture problem.
Example provided: motion direction of a square moving from one position (blue) to another (green).
Local motion detectors may perceive only general downward motion without the ability to discern left/right shifts accurately.
Global motion detector synthesizes these inputs to reveal overall motion direction (downward and right).
Structure of Motion Detection Across the Visual Pathway
Discussion on dorsal stream processing for motion that identifies where things are in space.
Lesions in the magnocellular layers of the LGN impair perception of large, rapidly moving objects.
Area MT (or V5) identified as a motion-sensitive region responsive to specific motion directions.
Stimulating area MT leads to perceptions of motion, emphasizing its role in motion processing.
Self Motion vs. Object Motion
Distinguishing Between Movements
Need to differentiate between self-motion (one's own movement) and motion of objects in the environment.
Reichardt detectors alone cannot make this distinction; require additional information.
Sources of Information for Distinction
Vestibular System: Tracks balance and head movement, providing crucial information on self-motion.
Efferent Copy: A signal sent from the motor cortex to indicate eye movement, helping to stabilize visual perception amidst self-motion.
Comparator Mechanism: Integrates retinal motion information and motor signals to deduce whether the motion perceived is due to self or external objects.
Eyeball Movement Dynamics
Moving eyes alters the position of objects across the retina without actual motion in the external environment (retinal movement signal).
Efferent copy signal aids in maintaining stable perception of the world while moving one's eye.
Optic Flow and Depth
Optic Flow
Describes how objects appear to shift in relation to a moving observer, illustrating depth perception.
Examples provided of visual patterns seen during movement, emphasizing the focus of expansion where the scene radiates outwards.
Depth Cues from Motion
Incorporation of optic flow into the understanding of depth as one moves through an environment.
Differences in radial expansion of nearer versus farther objects aid in deriving depth information.
Structure From Motion (SfM)
Definition
Structure from motion is the ability to perceive three-dimensional shapes from two-dimensional moving images.
Examples and Applications of SfM
Demonstrated ability to deduce the shape of objects, such as cylinders, from limited points of light.
Explains Rigid vs Non-Rigid Motion: Majority focus is on rigid structures (unchanging shape) but acknowledges non-rigid biological forms (like jellyfish) and their movement complexities.
Biological Motion Processing
Use of point-light displays to study human movement interpretation. Observers can identify actions (walking, kicking, throwing) despite minimal visual cues.
Depth Processing Introduction
Importance of Depth Processing
Critical for mobile organisms to interpret environments and navigate effectively.
Distinction made between depth (proximity without scale) and distance (specific measurement).
Challenges of Depth Processing
Real-world three-dimensional space is compressed into a two-dimensional retinal image.
Inverse optics: Brain's challenge of interpreting a 2D representation into a 3D understanding.
Depth Cues Types
Categorization of Depth Cues
Non-metric Depth Cues: Provide relative depth information.
Metric Depth Cues: Provide quantifiable distance details (e.g., proximity or relation).
Monocular vs Binocular Cues:
Monocular—function with one eye (e.g., pictorial cues).
Binocular—require both eyes (e.g., stereoscopic cues).
Kinematic Depth Cues
Motion Parallax: Displacement of objects based on the observer's movement; closer objects move faster across the retina.
Example: Utility poles in motion demonstrate speed differential based on distance.
Optical Expansion: Objects increase in size as they approach; an expanding image indicates that the object is getting closer.
Accretion and Deletion: Addition or removal of visual texture as objects move; important for discerning layered structures.
Rainier Perspectives: Gain sense of depth by observing how textures change as objects occlude or reveal each other during motion.
Preliminary Conclusion
Q&A emphasized to ensure understanding of motion detection and transition to depth processing, organizing lecture in a context that combines multiple aspects of two-dimensional and three-dimensional visualization.
Next Lecture Preview: Begin exploring binocular depth cues, particularly binocular disparity, and how this information allows for richer, metric depth perception from visual inputs.