Visual Perception: Surfaces, Objects, and Gestalt Principles

Surface

The 2D boundary between an object and whatever surrounds it.

Object

A bounded 3D volume of connected material.

Role of surfaces

Organize separate visual features into coherent units that can form objects.

Visual processing sequence

Features/fragments → surfaces → objects.

Grouping

Collecting visual elements that appear to belong together.

Purpose of grouping

First step in discovering surfaces and then objects.

Gestalt laws of grouping

Rules describing which visual elements are likely to belong to the same object.

Proximity

Elements close together tend to group together.

Similarity

Elements that look similar tend to group together.

Closure

Elements enclosing a region tend to group together.

Why closure helps object perception

Objects are bounded volumes and therefore have closed outer contours.

Good continuation

Elements forming the smoothest continuous contour tend to group together.

Good form

Contour fragments forming the simplest, most regular, or familiar shape tend to group.

Bad form

Cues indicating that contours or surfaces probably do not belong together.

Two important bad-form cues

T-junctions and X-junctions.

T-junction

A junction where one contour terminates against another contour.

What a T-junction indicates

Occlusion: one surface is in front of another.

Stem of a T-junction

Contour belonging to the surface lying behind.

Top of a T-junction

Contour belonging to the surface lying in front.

Why the back contour stops at a T-junction

The opaque front surface occludes it.

X-junction

Independent overlapping contours, often caused by a transparent front surface.

T-junction vs. X-junction

T = one contour interrupted by occlusion. X = both contours continue through overlap.

Transparent surface effect on junctions

A T-junction can become an X-junction when the front surface is transparent.

Modal and amodal completion

Ways the visual system adds surfaces or contours beyond what is physically present.

Amodal completion

A surface is represented as continuing behind an occluding surface.

Visibility in amodal completion

The hidden part is not seen but is registered as continuing.

Cues for amodal completion

Similarity and good continuation.

Example of amodal completion

Seeing one complete face behind bars rather than disconnected face fragments.

Stereopsis

Depth perception produced from differences between the two eyes' images.

Stereopsis and amodal completion

Depth establishes front vs. back surfaces and improves recognition of hidden objects.

How amodal completion helps recognition

It groups separated visible fragments into one partially hidden object.

Modal completion

A missing contour or surface in front is actually perceived.

Subjective contour

A perceived contour where no physical contour exists in the image.

Condition for modal completion

The surfaces have the same color and brightness.

More common type of completion

Amodal completion.

Modal completion preference

The visual system tends to choose the interpretation requiring the shortest modal contours.

Kanizsa illusion

Demonstrates both modal and amodal completion.

Modal completion in a Kanizsa square

You perceive the square's boundary even though no physical boundary exists.

Amodal completion in a Kanizsa square

Interrupted discs are perceived as complete behind the illusory square.

What weakens subjective surfaces

Changes that make alternative interpretations more plausible.

Subjective surfaces

Show that the visual system can invent surfaces not explicitly present in the retinal image.

Figure-ground organization

Determining which surface is foreground and which is background.

Border ownership

Determining which neighboring surface a shared border belongs to.

Which surface owns a border

The surface perceived as being in front.

Occlusion border and the back surface

The border does not end the back surface because it can continue behind the front surface.

T-junctions and border ownership

T-junctions indicate which surface is in front and therefore owns the border.

Good form and border ownership

The visual system favors interpretations that produce simpler and more regular surfaces.

Border ownership dispute

An ambiguous image where the same border can be assigned to different surfaces.

Why ambiguous figures can switch

Border ownership changes, altering which surface appears in front.

Top-down object knowledge

Prior knowledge about objects influences how visual fragments are grouped and recognized.

Object knowledge and grouping

A few distinctive features suggest an object and guide how remaining fragments are joined.

Face behind bars

Grouping, amodal completion, and object knowledge make us perceive one complete face.

Limit of object knowledge

Surface continuity can override object knowledge.

Grouping vs. completion

Grouping joins related visible elements, while completion represents missing contours or surfaces.

Modal vs. amodal completion

Modal = missing contour is seen. Amodal = hidden continuation is registered but not seen.

Easy T vs. X rule

T = one contour terminates → occlusion. X = both continue → transparency or overlap.

Lecture 7 hierarchy

Bits → surfaces → objects.