Chapter 5: Perceiving Objects and Scenes
● The perceptions we make as humans come naturally to us w/o much effort however
the case is not the same for computers/machines
Why Is It So Difficult to Design a Perceiving Machine?
● Descriptions of difficulties involved in designing a “perceiving machine”
● Humans can solve these problems easily however computer vision systems face
difficulty in doing so
The Stimulus on the Receptors Is Ambiguous
● Extending rays from the object into the eye determines its image on the retina
● Inverse projection problem: the task of determining the object that created the
image on the retina
○ This task is where the visual system’s job starts
○ “Inverse” because it starts with the retinal image and extending rays out of the
eye
○ This problem is hard to solve because a particular image on the retina can be
created by many diff objects ⇒ image on retina is ambiguous
● Incorrect perceptions don't occur most of the time since our visual system solves the
inverse projection problem and determines which object was responsible for a
particular image on the retina out of all the possibilities
Objects Can Be Hidden or Blurred
● People can easily perceive objects that are partially hidden by other objects as whole
○ they understand that the part that is covered still exists
○ they use their knowledge about the environment to determine what that object
could be
● The problem of hidden objects occurs frequently in the environment but it's easily
solved by humans
● People are able to recognize blurred objects
Objects Looks Different From Different Viewpoints
● The images of objects are constantly changing depending on which angle they are
being viewed from
, ● Viewpoint invariance: the ability to recognize an object seen from different
viewpoints
○ This ability is found in humans
○ It allows us to be able to tell whether a face seen from different angles is the
same person or not
● Conclusion:
○ The problems listed above that pose a challenge to perceiving machines prove
that the process of perception is complex
○ Considering “perceptual organization” will help us understand how humans
overcome these complexities
Perceptual Organization
● Perceptual organization: the process by which elements in the environment become
perceptually grouped to create our perception of objects
● Process of perceptual organization has 2 components:
○ Grouping: the process of “putting together” or organizing stimuli into units or
objects/to create a coherent image of a thing
■ Example: seeing the Dalmatian dog in figure 5.12 because we have
perceptually grouped some of the dark areas to form a Dalmatian
○ Segregation: the process of separating an object/area from another
object/area
■ Example: seeing two buildings in a city scene as separate from e/o and
recognizing their borders
○ Grouping and segregation work in conjunction
The Gestalt Approach to Perceptual Grouping
● Gestalt psychologists: Gestalt means configuration (rough translation)
○ They asked the Q “How are configurations formed from smaller elements?”
Structuralism
● Structuralism:
○ Approach that came before the Gestalt approach
○ Proposed by Wilhem Wundt
○ Distinguished between sensations (simple processes that occur due to
stimulation of the senses) and perceptions (more complex conscious
experiences such as our awareness of objects)
● The perceptions we make as humans come naturally to us w/o much effort however
the case is not the same for computers/machines
Why Is It So Difficult to Design a Perceiving Machine?
● Descriptions of difficulties involved in designing a “perceiving machine”
● Humans can solve these problems easily however computer vision systems face
difficulty in doing so
The Stimulus on the Receptors Is Ambiguous
● Extending rays from the object into the eye determines its image on the retina
● Inverse projection problem: the task of determining the object that created the
image on the retina
○ This task is where the visual system’s job starts
○ “Inverse” because it starts with the retinal image and extending rays out of the
eye
○ This problem is hard to solve because a particular image on the retina can be
created by many diff objects ⇒ image on retina is ambiguous
● Incorrect perceptions don't occur most of the time since our visual system solves the
inverse projection problem and determines which object was responsible for a
particular image on the retina out of all the possibilities
Objects Can Be Hidden or Blurred
● People can easily perceive objects that are partially hidden by other objects as whole
○ they understand that the part that is covered still exists
○ they use their knowledge about the environment to determine what that object
could be
● The problem of hidden objects occurs frequently in the environment but it's easily
solved by humans
● People are able to recognize blurred objects
Objects Looks Different From Different Viewpoints
● The images of objects are constantly changing depending on which angle they are
being viewed from
, ● Viewpoint invariance: the ability to recognize an object seen from different
viewpoints
○ This ability is found in humans
○ It allows us to be able to tell whether a face seen from different angles is the
same person or not
● Conclusion:
○ The problems listed above that pose a challenge to perceiving machines prove
that the process of perception is complex
○ Considering “perceptual organization” will help us understand how humans
overcome these complexities
Perceptual Organization
● Perceptual organization: the process by which elements in the environment become
perceptually grouped to create our perception of objects
● Process of perceptual organization has 2 components:
○ Grouping: the process of “putting together” or organizing stimuli into units or
objects/to create a coherent image of a thing
■ Example: seeing the Dalmatian dog in figure 5.12 because we have
perceptually grouped some of the dark areas to form a Dalmatian
○ Segregation: the process of separating an object/area from another
object/area
■ Example: seeing two buildings in a city scene as separate from e/o and
recognizing their borders
○ Grouping and segregation work in conjunction
The Gestalt Approach to Perceptual Grouping
● Gestalt psychologists: Gestalt means configuration (rough translation)
○ They asked the Q “How are configurations formed from smaller elements?”
Structuralism
● Structuralism:
○ Approach that came before the Gestalt approach
○ Proposed by Wilhem Wundt
○ Distinguished between sensations (simple processes that occur due to
stimulation of the senses) and perceptions (more complex conscious
experiences such as our awareness of objects)
