Image Processing Reference
In-Depth Information
construction direction that combines low computational cost with both high precision and
high resolution. Research studies have shown [ 1 ] that the system based on digital fringe pro-
jection has acceptable performance, and moreover, it is insensitive to ambient light.
The major stages of fringe projection method are:
• A sinusoidal patern is projected over the surface of the object by using a projector, as illus-
trated in Figure 1 .
FIGURE 1 Fringe projection arrangement from [ 2 ].
• A digital camera is used to capture the patern that has been phase modulated by the topo-
graphy of the object surface.
• The captured patern is analyzed to extract topographical information.
• A phase-unwrapping algorithm is executed to resolve discontinuities on the extracted
phase.
• A phase to height conversion method is used to materialize the third dimension in every
pixel.
Practical considerations regarding the fringe projection system in general and the binary
code unwrapping method in particular are the focus of this paper. Section 2 highlights the pre-
vious research into fringe analysis. In Section 3 , practical issues for fringe patern generation
are described. Section 4 describes the details of binary code generation for phase ambiguity
resolution and evaluation of this unwrapping method. In Section 5 , practical issues for fringe
pattern photography are described. Section 6 offers concluding remarks and suggestions for
future work.
2 Prior and related work
There have been many advances regarding structured light systems for 3D shape measure-
ment such as multiple level coding, binary coding, triangular coding, and trapezoidal coding.
Fringe projection 3D shape measurement improves upon the usual structured light systems
because high-resolution (as high as camera resolution and projector) results in a correspond-
ingly high speed. Furthermore, phase of each pixel, which includes modulated depth inform-
ation, is calculated through gray-level intensities. The captured phase can be transformed to
height (depth) for each pixel after discontinuities in the phase have been resolved by phase
unwrapping.
Phase detection approaches have been based on a variety of approaches such as Fourier
transform [ 3 - 6 , 14 ] , interpolated Fourier transform [ 7 ] , continuous wavelet transform [ 8 - 10 , 15 ] ,
2D continuous wavelet transform, discrete consign transform, neural network [ 11 ] , phase
locked loop, spatial phase detection, and phase transition.
 
 
Search WWH ::




Custom Search