Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
2~333G
FIELD OF THE INVENTION
The invention relates to three dimensional (3-D) color
imaging of the profile of a target surface.
BACKGROUND OF THE INVENTION
Monochromatic 3-D imaging of objects is known and has
applicability to the accumulation of detailed data on the
shapes and surface profiles of objects (articles, scenes
and/or persons), to the automatic inspection or assembly of
o~jects, to robotics generally, and to various medical
applications.
PRIOR ART
It has been known for many years that optical
triangulation can yield accurate knowledge of range and hence
- of the profile of a target surface. Typical prior U.S.
patents that describe implementation of the triangulation
principle are 3,986,774 (Lowrey et al); 4,171,917, oct. 23,
1979 (Pirlet); 4,349,277, Sept. 14, 1982 (Mundy et al);
4,627,734, Dec. 9, 1986 tRioux); and 4,701,049, Oct. 20, 1987
(Bec~man et al).
The patents to Pirlet and Rioux teach triangulation
configurations in which the surface is scanned by a beam of
light. A synchronously scanning receiver images reflected
light onto a position sensitive detector, e.g. a CCD (charge
coupled device), to generate electrical signals indicative of
range deviations of points on the surface from a reference
plane.
Beckman et al also disclose a measuring system employing
the triangulation principle. This patent is directed to
techniques for improving resolution by varying the cross-
section of the measuring beam, and includes a feature of
viewing a lighted dot on the target surface at two different
angles to discriminate a true reflection from a false one.
Mundy et al employ the optical parallax triangulation
principle in which a color pattern is projected onto the
surface, shifts of wavelength bands being detected on separate
~~ detector arrays, these shifts corresponding to the profile of
the surface.
U.S. Patent No. 4,645,347, Feb. 24, 1987 (Rioux) teaches
another method of measuring profile. It uses a converging
lens with a mask having two apertures. The spacing between
images on a detector represents the range deviation of points
on the target surface from a reference plane, e.g. the focal
plane of the converging lens.
Alternatively, the range data can be detected by methods
other than the triangulation method, such as by time of flight
(radar) measurement. A full summary of the various methods of
optical ranging is provided in "Active, Optical Range Imaging
Sensors" by Paul J. Besl, published in Machine Vision and
Applications (1988) 1:127-152.
However, none of these known systems also collects data
--- on the color of the target surface.
On the other hand, there are many prior patents on color
video cameras. U.S. Patent No. 4,072,405, Feb. 7, 1978
(Ozeki), for example, uses a prism combination together with
color filters in a video camera to separate the three primary
color components. Three detectors, one for each primary
color, produce the red, green and blue signals, which are
processed into a color TV signal. Total color separation is
2S required, thus calling for separate detectors.
SUMMARY OF THE INVENTION
The primary objective of the present invention is to
provide a method by which the advantages of collecting color
data along with the 3-D data can be achieved in a way that is
so simple that it requires only relatively minor modifications
to the known monochromatic 3-D cameras already in use.
More specifically, the objective is to avoid or minimize
the complications of multiple detectorsj or multiple color
filters, or other major changes to the components required in
a 3-D camera.
The invention meets this requirement by providing a
2~333~
method and apparatus in which the target surface is scanned by
~~ an incident beam of light that contains a plurality of
wavelengths including at least one well defined wavelength.
A return beam reflected from the surface is dispersed into a
plurality of return beams which are received by a position
sensitive detector to generate signals that represent the
color of the surface.
In this system only a single position sensitive detector
is needed. Moreover, this detector can function mono-
chromatically, i.e. without any need for color filters or
other color sensitive devices.
As a result, the invention enables existing,
monochromatic, 3-D cameras to be readily modified to furnish
color data. It is only necessary to change the nature of the
lS source of light, to provide a color dispersing device in the
return beam, and to modify the processing of the output
signals from the detector in the camera. In all other
respects, the camera can remain unchanged.
In one form of the invention the incident beam comprises
three well-defined wavelengths, namely those of the primary
colors, blue, green and red. This form of the invention is
convenient for recording digital color data along with the
digital profile data and for generating from this data
displays suitable for viewing with the human eye.
- 25 In another form of the invention the incident beam mixesone well defined wavelength with a continuum of visible light,
the well defined wavelength preferably being outside the
visible spectrum, e.g. infrared, in order to be clearly
distinguishable therefrom. This form of the invention is well
adapted to applications in which very precise color
measurements (colorimetry) are required, since the signals
received in the detector from the visible continuum portion of
the beam can be broken down into a relatively large number of
wavelengths for detailed analysis.
4 2~4~ ~ 6
BRIEF DESCRIPTION OF THE DRAWINGS
~~ Figure 1 is a schematic illustration of a first
embodiment;
Figure 2 shows signals generated in the embodiment of
Figure 1;
Figure 3 shows a second embodiment;
Figure 4 is a modification of Figure 1;
Figure 5 shows the signals generated by the arrangement
of Figure 4;
Figure 6 shows a further embodiment;
Figure 7 shows signals generated in the embodiment of
Figure 6;
~ igure 8 is a modification of Figure 6;
Figure 9 shows signals generated in the arrangement of
Figure 8; and
Figure 10 shows a fragment of a modification of the
system of Figure 1.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Figure 1 shows schematically a synchronous optical
triangulation scanning system that functions in accordance
with the teachings of Rioux U.S. Patent 4,627,734 and is
essentially alike in structure to the embodiment illustrated
in Figure 12 of such patent.
-' A light source 2, e.g. a RGB laser, produces a beam 6
that contains well defined red, green and blue wavelengths.
Together with fixed mirrors 10, one surface of an oscillating
double-sided mirror 4 scans the beam 6 in the X direction and
projects it towards an object 8. Light 7 received back from a
point P on the target surface of the ob~ect 8 is returned by a
further fixed mirror 10, the opposite surface of the mirror 4,
and a lens 14, in the form of a return beam 12 that is imaged
onto a position sensitive detector 18, e.g. a CCD. Interposed
in this beam 12 is a device 16 for dispersing the beam into
separate return beams 12B, 12G and 12R of the three primary
colors. While the dispersal device 16 can be a simple wedge,
it is preferable to use either a double wedge or other device
5 ~a433~6
that will achieve a collinear effect, at least for one of the
~ beams, preferably the green beam. In other words, the beam
12G will be a straight through continuation of the beam 12.
This collinearity is, however, not essential.
The detector 18 measures the amplitudes A and the
positions D of the respective beams 12B, 12G and 12R to
generate the signals 17B, 17G and 17R shown in Figure 2. The
position of any one of these signals indicates the range of
the point P, i.e. the deviation of the point P in the Z
direction from a reference plane Z = 0, such plane being
perpendicular to the optical axis of the beam 12. The
detector 18 is slanted to this optical axis because the focal
plane ~aries with range. Since the positions of the 17R, 17G
and 17B signals relative to each other do not vary
substantially, any one, two, or aIl of these signals can be
used to measure the Z deviation. Usually the signal with the
~- greatest amplitude will be chosen for this purpose. If the
color of the object is such that one of these signals is
absent or is too small to measure, the colors of the~two
remaining signals can be identified by their spacing from each
other.
If the detector 18 is two-dimensional, i.e. has pixels
extending in the Y direction (perpendicular to both the X and
Z directions) as well as in the X direction, the necessary
" 25 scanning of the beams in the Y direction can conveniently beachieved by means of a further mirror oscillating about an
axis extending in the X direction (see the mirror M6 in the
,~
Figure 12 of U.S. 4,627,734 referred to above).
While oscillating mirrors are the preferred method of
scanning, it is possible to achieve the same result by
relative translation of the object 8 and the entire 3-D camera
19 that the other components so far described in Figure 1
represent.
As in the prior art, a microprocessor 20 controls by a
connection 13 the scanning of the mirror 4 (or the mirrors, if
there are more than one), or the relative translation of the
object 8 and the camera 19, while receiving and digitizing
6 2~4~6
signals 17B, 17G, 17R on line 15 to produce 3-D color data
~ that can be stored, used to identify the object or used to
create a facsimile of it, which data can also be shown on a
display 21.
A more basic triangulation configuration is shown in
~igure 3 wherein a beam 32 from an RGB laser, which beam is
scanned in the Y direction by a mechanism (not shown), is
projected onto a surface point P of an object 34. After
passing through an imaging lens 35 and a dispersing device 38
for splitting the return beam 33 into the three primary colors
(not separately shown in Figure 3), such beam is received at M
on a linear position sensitive detector 36. As before, the
shift Q of the point M from a reference point N for a beam 33'
that would be received from the reference plane Z = 0 in the
absence of the object 34, represents the height H of the point
P from this plane. As before the color of the surface can be
measured from the relative intensities of the three color
beams. B is the triangulation angle and the slant of the
detector 36 is a straight line projection from a point on the
incident beam 32 that is intersected by an extension of the
central transverse plane of the lens 32, as shown by the
dotted lines.
Figure 4 shows an embodiment similar to Figure l except
that the light source 22 produces a beam 6' that is a mixture
of a continuum of visible light and laser light of a well
defined wavelength, preferably a wavelength, such as infrared,
outside the visible spectrum. Alternatively, the wavelength
of the laser light can be within the visible spectrum,
provided the resulting signal is distinguishable from that
resultin~ from the visible light. The corresponding separate
return beams 12V and 12I received by the detector 18 now
generate respective signals 17V and 17I. The signal 17I is
used to determine the deviation of the point P in the Z
direction while the signal 17V can be broken down into a
relatively large number of wavelengths by the microprocessor
20 to enable a much more accurate colorimetric measurement
than can be achieved simply from the three primary colors.
~ 2~ 4~33B
This system thus makes possible the detection of minor or other
subtle color changes of a surface over time or during processing of
such surface, e.g. cleaning or otherwise restoring.
Figure 6 shows, in simplified form, an embodiment employing the
optical configuration shown in Rioux U.S. patent 4,645,347. A laser
48 generates a so-called lamina or planar, multiwavelength beam 49,
i.e. a beam that has a substantial extent in the Y direction while
0 being very narrow in the X direction, whereby to project a stripe of
light onto the surface of the object 8. By virtue of an apertured
mask 44, a lens 42 focuses light 45 from the target onto a pair of
spaced points on a two-dimensional, position sensitive detector 52,
e.g. a CCD array, the spacing between these points being a measure of
the Z deviation of the point P from the reference plane. Because the
beams 45 pass through color dispersing device 50, each beam is split,
as before, into beams 45B, 45G and 45R, the signals seen by the
microprocessor 20 being as shown by the three pairs of wavy lines
62B, 62G and 62R in Figure 7. The amplitudes of the lines are
processed as before to generate data on the color of the target
surface, while the spacings Dl, D2, etc. at different locations in
the Y direction between a selected pair of lines, say the pair of
lines 62G, indicates the Z data, i.e. the deviation in range. As in
the previous examples, the apparatus will include means for scanning
the light beam relative to the object in the X direction.
Figure 8 shows an embodiment similar to that of Figure 6 and
but modified to employ a light source 70 for generating a laminar
beam 71 that combines the visible continuum with a well defined laser
wavelength as in figure 4. As a result the detector 52 receives
beams 72X and 72I that have the shapes shown in Figure 9. As in
Figure 7, spacings Dl, D2 provide that Z data, while the nature of
the spectrum 74 indicates the color of the surface.
Figure 10 shows a modification in which the triangulation
method of measuring range and hence the profile of the target
~.,
., ,.ff
8 2~333~
surface is replaced by a time of flight (radar) type of
~- system. A partial mirror 80 deflects a part of one of the
return beams, say the beam 12G, into a detector 81 connected
to the microprocessor 20. The incident beam is pulsed and the
microprocessor measures and digitizes the range on the basis
of the detected time of flight of the pulse from emission to
return. The detector 18 need no longer be slanted to the
~- optical axis, since it is no longer being used to measure
range (profile), but it has been shown slanted in Figure 10,
because this is its usual orientation in a 3-D camera and one
of the features of the present invention is a minimization of
changes to the camera. Since the detector 18 is no longer
required to be sensitive to range, the triangulation angle can
be reduced to approximately zero, i.e. the incident beam can
have an axis approximately parallel to that of the return
' beam, which is advantageous, especially when the target
-~ surface is comparatively remote from the camera. Hence the
Figure 10 embodiment, while calling for more modification to
the camera than the methods of
Figures 1-9 by virtue of the need for an additional detector
- 81, has the offsetting advantage of permltting adoption of a
zero or very small angle between the axes of the incident and
return beams.