Introduction vision

Introduction

to

Computer Vision

[email protected]

goals of field of vision

• understand how animals represent and process information carried by light, by– measuring and modeling visual performance

in humans and other animals– finding ways to build artificial visual systems– characterizing neural mechanisms that

• implement visual systems– apply this understanding to obtain medical,

technological advances

processing of images in humans

• as a first approximation, rods and cones (sensory cells in the retina) represent image as large 2D array of light intensities– about 126 million sensory cells!

• this image representation is processed by brain enabling complex cognitive functions– recognize a familiar face or scene– disambiguate overlapping objects– read sloppy handwriting

• how does the brain do all of this? how might image processing be partitioned into subtasks?

image processing tasks of brain

• possible tasks:– extraction of contour (e.g. sharp light intensity

changes in the image)– extraction of motion– identification of object parts

• still unclear: how are these integrated to enable us to extract meaning from what we see?

psychophysical experiments

• used to test hypotheses about how the brain/mind processes optical information

Examples

• Inattentional Blindness

• Change Blindness

• Figure-Ground Segregation

Inattentional Blindness

Mack & Rock (1998)• Definition: the failure to see consciously, caused by

lack of attention• We can miss perceiving very obvious changes if we

are not attending. Subjects do not consciously perceive features of the visual scene that they do not attend to.

• Subjects were engaged in tasks that demanded a high degree of attention, such as looking at a cross and trying to determine which arm is longer.

Trials 1

Trials 2

Trials 3

Inattention Trial

Recognition Test

Inattentional Blindness

• 25% of subjects failed to see the square when it was presented in the parafovea (2° from fixation).

• But 65% failed to see it when it was at fixation!

• What is missed?

– Sad or Neutral face

– A word (priming for the word is present)

• What is not missed?

– Name

– Smiling face

Change Blindness

• Change Blindness is the phenomena were we fail to perceive large changes, in our surroundings as well as in experimental conditions.

• Change could be in existence, properties, semantic identity and spatial layout.

• Attention is required to perceive change, and in the absence of localized motion signals, attention is directed by high level of interest (Rensink et al, 1997).

Flicker paradigm

• Basically, alternate an original image A with a modified image A’, with brief blank fields placed between successive images

Change Blindness

• “Visual perception of change in an object occurs only when that object is given focused attention”

• “In the absence of such attention, the contents of visual memory are simply overwritten by subsequent stimuli, and so cannot be used to make comparisons”

CB – Another Example

Why CB?

• Change blindness could be due to –– Poor representation of pre- and post change

scene or – Pre change representation gets over-written

by post change representation or– Capacity to retrieve and compare information

is limited (Hollingworth, 2003).

• Color change detection with multi-element displays

Figure-Ground Segregation

• Discovered by Edgar Rubin (Fig.1, 1921).

• Only one side of the contour is seen as figure.

• Has shape, appears closer.

• Background appears behind the figure and has no shape.

Fig.1 (Faces/vase display)Fig.1 (Faces/vase display)

Background• Configural cues:

– symmetry– convexity – area ………….. Gestalt psychologists

• Lower region • lower region of a display will be seen as figure than the

upper region. (Vecera et al. 2002)

• Top-bottom polarity• Stimuli having wide base and narrow top were perceived as

figures than the ones which had narrow base and wide top.• (Hulleman and Humphreys, 2004)• Higher cognitive processes

» Object memory (Mary Peterson et al. 1991)» Attention ( Vecera et al. 2004)

Motivation

• Palmer and his colleagues conducted a study in which they used temporal frequency (flicker) and manipulated edge synchrony with the two regions (left and right).

• They concluded from their studies that edge plays the key role in determining figure-ground segregation. In their study they found that the region with which the edge synchronizes will be seen as figure irrespective of whether the region is flickering or not.

• Wong and Weisstein (1987) demonstrated that spatial and temporal frequencies play a major role in assigning figural status to a region.

Reference Books

• Fundamentals of Digital Image Processing: Anil K. Jain

• Digital Image Processing: Gonzalez & Woods • The Image Processing Handbook: J.C. Russ • Digital Image Processing: B. Jahne

• Image Processing, Analysis and Machine Vision: M. Sonka, V. Hlavac, R. Boyle

• Computer Vision Handbook: B. Jahne • Computer Vision: M. Brady