• Binocular stereo

(1)

Chia-Yen Chen CS716 - 3D Shape Recovery 1

Shape recovery methods

This set of notes contains drawing for the following methods:

• Binocular stereo

• Structured lighting

• Shape from contours

a Binocular stereo b

(2)

a

b

Disparity for b Binocular stereo

Left camera Right camera

Disparity for b Disparity for a Compare

Disparity for a > Disparity for b Means ais closer than b Binocular stereo

(3)

a b

Note the order of the points’ projections on the imaging plane Binocular stereo

Left camera Right camera

a Structured lighting b

(4)

Structured lighting

Camera Light source

Calibration plane

Projected light d₀

d₁

x₀ x₁

Structured lighting

x d

x₀ d₀

d₁

x₁

Graph of projected light

Function of projected light:

d = F(x) = mx+c

(5)

Shape from contours

Segmentation and extraction of contours

Shape from contours

(6)

Shape from contours

A preliminary 3D model (visual hull) and a horizontal cross section of the model

Shape from contours

(7)

Shape from contours

Consider this horizontal cross section of an object. Can this shape be accurately recovered using shape from contours?

Assume: object is placed on a turntable and rotated about the vertical axis

camera

Cavity cannot be recovered using SFC