It is simpler than conventional stereo, only requires a single
camera, and provides more detailed information.
It is faster than laser scanning techniques.
The equipment is cheap and readily available.
The extraction and use of intrinsic surface information mean
that classification and defect detection are more robust and
accurate.
It can be made surface rotation invariant.
The surface relief and any printed pattern may be separated
and analysed independently.
It provides a cheap method of capturing bump and colour
maps of 3D surface textures for augmented reality applications.
What are
the requirements and restrictions?
Illumination: controlled illumination is required.
Scale: the scale of the target purely depends on the limitations of
optics and illumination: it can therefore range from the microscopic
up to any target for which the illumination can be controlled.
Resolution: The spatial resolution is limited by the camera technology,
currently 24 bit colour and 2-3 million pixels (but this is doubling
every 2 years).
Images - Photometric texture analysis requires three registered images
of the target taken under 3 different illumination conditions.
Alternatively, for monochrome targets, a single colour coded image can be captured.