SPATIAL PROCESSING AND DEPTH PERCEPTION

SPATIAL PROCESSING AND DEPTH PERCEPTION

Problem Statement

  • Problem: How do we reconstruct a three-dimensional world from distorted and curved 2-dimensional retinal images?

  • Answer: We utilize various monocular and binocular cues to achieve this reconstruction.

Introduction to Depth Cues

  • Depth cue: Information that provides insight into the third dimension (depth) of visual space.

    • Monocular depth cue: A depth cue that is accessible even when the world is viewed with one eye alone.

    • Binocular depth cue: A depth cue that necessitates information from both eyes for perception.

Key Definitions

  • Binocular disparity: The differences observed between the two retinal images of the same scene.

    • Stereopsis: A vivid perception of the three-dimensionality of the world, which arises from binocular disparity and is absent in monocular vision.

Binocular Depth Cues

  • Disparity in the Retinal Images:

    • Human sensitivity to small binocular disparities is remarkable; some adults can detect a difference in depth as small as 4 mm at a distance of 5 m.

    • Stereopsis typically provides information about relative depth up to a distance of about 200 m. This functional capacity offers significant adaptive and evolutionary advantages to individuals.

Monocular Cues to Three-Dimensional Space

  • Occlusion: A cue representing relative depth order where one object partially obscures another object, indicating that the obscured object is farther away.

  • Relative Size: This cue involves comparing the sizes of various objects without knowledge of their absolute sizes.

    • Generally, we assume that smaller objects are located farther away from the observer than larger objects, all else being equal.

  • Relative Height: For objects that are touching the ground:

    • Objects appearing higher in the visual field are perceived to be farther away.

    • Conversely, in the sky above the horizon, objects lower in the visual field appear to be farther away.

  • Familiar Size: This cue is based on the inherent knowledge of the typical size of various objects.

    • Knowing what size an object typically is helps create estimations about its distance based on how small or large it appears in the visual field.

  • Aerial Perspective: A depth cue that leverages the understanding that light gets scattered by the atmosphere.

    • More atmosphere results in additional light scattering; thus, the more distant an object is, the fainter, bluer, and less distinct it appears.

  • Dynamic Cues:

    • Motion Parallax: A phenomenon where images closer to the observer move more rapidly across the visual field compared to images that are further away.

    • The brain utilizes this motion information to deduce distances of objects in the environment.