Effect of observer translation on a small patch of the image

From A conversation about the brain
Jump to: navigation, search
Translation of the observer causes quite simple changes in a small patch of the image. If the surface is planar and slanted in depth, the direction of 'squashing' or sheer in the image indicates the direction of the observer's movement (horizontal here); the line that is unchanged indicates the direction of tilt (i.e. the axis of the slant, vertical here); and the amount of squashing or sheer is proportional to the degree of slant in depth.
Translation of the eye/camera changes the optic array. For a small patch, this gives information about the magnitude and direction of slant of the surface (see Fig). Taken together with the information about relative visual direction, this allows a binocular observer, or a monocular observer translating small distances, to fixate surfaces and make judgements about their tilt and slant relative to the line of sight [1]. The information stored could contribute to what Marr described as a 2$\frac{1}{2}$D sketch [2], i.e. one that is quite closely related to retinal images and yet it survives eye movements and small head movements.

Back to 3D vision.

References

  1. Glennerster, A. (2013). Representing 3D shape and location. In Shape Perception in Human and Computer Vision (pp. 201-212). Springer London
  2. Marr, D., & Nishihara, H. K. (1978). Representation and recognition of the spatial organization of three-dimensional shapes. Proceedings of the Royal Society of London. Series B. Biological Sciences, 200(1140), 269-294.