Practical issues for binary code pattern unwrapping in fringe projection method - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 477

3 Practical issues for fringe pattern generation
Fringe patern generation is the process of projecting a sinusoidal patern over the surface of
an object and capturing the patern by taking a picture of it with a digital camera. The object
used for illustration is a chalky white sculpture of a woman without any sudden change in
depth in its topography as shown in Figure 2 .
FIGURE 2 Object used for experimentation.
In our previous work [ 1 ] , an experimental study and implementation of a simple fringe
projection system have been reported that was based on a multiwavelength unwrapping ap-
proach. In that work, a classification of 3D reconstruction systems was also presented. In
later work [ 12 ] , a new method of phase unwrapping was implemented that is based on time
analyses approaches. The fringe projection system provided an experimental environment in
which the two unwrapping methods could be compared. Experimental results have shown
that the digital code patern unwrapping method is a stable and reliable method that results
in a higher level of precision in the reconstructed 3D model at the cost of using more pictures.
Using the same object to evaluate each unwrapping technique, it was found that binary code
pattern unwrapping resulted in a more accurate point cloud.
Matlab high-level functions were used to produce the processing software to analyze the
images and create the 3D models. Although there are other available choices of programming
languages for image processing, Matlab has unique advantages for working in this field. In
fact, we have created a magnificently powerful yet simple tool by taking advantage of Matlab's
image processing capabilities. Processing functions will be described by showing fragments of
Matlab code.
In the fringe patern generation algorithm, creating the fringe paterns for one row will re-
peat for all the similar rows. Therefore, only one row of the algorithm will be explained. As-
sume that the projector has 100 width pixels. Assume that one fringe patern is to be created
with wavelength of 20 (T = 20), wave frequency of ω, three fringe shifts (N = 3), fringe shifting
of 120°, and shifted phase value of p 0 . Accordingly, the phase P = ωt + p 0 is a function of the
pixel position (t = 1:n). The wave frequency is ω = 2π/T. Therefore, a 0 + (b 0 -a 0 )*(1 + cos(P))/2 ex-
presses a sinusoidal function with values between a 0 and b 0 .
The brightness level of the fringe paterns in image pictures converts nonlinearly with the
level of brightness in the projected paterns due to the radiometric behavior of the projector-
camera system ( Figure 3 ) .