Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
1998-11-30
2003-05-27
Johnson, Timothy M. (Department: 2625)
Image analysis
Applications
Manufacturing or product inspection
C382S146000, C382S151000, C382S199000
Reexamination Certificate
active
06571006
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to machine vision, and particularly to dimensional measurement within a digital image.
BACKGROUND
Many machine vision systems use a model for inspecting or recognizing objects within images. Often, such models must accurately reflect the physical dimensions of the objects so that the position of the objects can be determined in an image, and so that precise tolerance checking and defect detection can be performed. For example, a vision-based automatic surface mounter (SMD) machine employs a model that includes the length of leads of leaded devices to accurately inspect and place the leaded devices on a printed circuit board (PCB).
FIG. 1A
illustrates a bottom view and a side view of a gullwing-leaded device
100
, not drawn to scale, where a leaded device is an electronic device that has a device body
110
and leads
102
. The leads
102
are metal contacts on the exterior surface of the device body
110
that are connected to an integrated circuit (not shown) within the device body
110
. Leaded devices include surface-mount devices and through-hole devices, for example. The leads
102
of the surface-mount devices are placed by the SMD machine, such that the leads
102
substantially contacts pads on a PCB (not shown) within positional tolerances.
The length
104
of a gullwing lead
102
is the distance between the base
108
and the tip
106
of the lead
102
. The positions of bases
108
and tips
106
in an image are often determined by identifying edges corresponding to the bases
108
and the tips
106
. With varying degrees of success, the edges are found using methods known in the art. The edges can also be found by using CALIPER TOOL sold by Cognex Corporation. The CALIPER TOOL is a machine-vision tool illustrated with reference to FIG.
2
and further described in Vision Tools, Chapter 4, CALIPER TOOL, Cognex Corporation, Version 7.4, 1996, pp. 208-231, incorporated herein by reference.
The CALIPER TOOL finds edges, such as
206
and
208
, within an area of an image
200
enclosed by a window
204
. More particularly, the CALIPER TOOL accumulates edges along a projection axis, p, of the window
204
. An edge, as used herein, consists of a plurality of connected edge elements or edge pixels that correspond to underlying image pixels. An image can be represented as an array of image pixels, where an image pixel is a picture element characterized by a grey value. Each edge can be one or more image pixels wide.
The intensity of pixels within the window
204
along p are projected (i.e., added), thereby generating a one-dimensional image
210
. The projection axis, p, is perpendicular to l, and together l, p, and w, which is the width of the window
204
, define the window
204
. Linear projection collapses an image by summing the grey values of the pixels in the direction of the projection. The summation tends to amplify edges in the same direction as p. After projection, an edge filter is applied to the one-dimensional image
210
to further enhance edge information and to smooth the one-dimensional image
210
. The one-dimensional image is illustrated graphically as histogram
212
. The edge
206
is represented in the histogram
212
as a falling ramp
216
, and the edge
208
is represented in the histogram as a rising ramp
218
. Each edge
206
and
208
has a polarity (i.e., direction), where edge
206
has a light-to-dark polarity and edge
208
has a dark-to-light polarity. In this example, both edges
206
and
208
have the same contrast, where contrast is the difference in grey levels on opposite sides of an edge.
The edges corresponding to the bases
108
and the tips
106
of the leads
102
are located using a window of the CALIPER TOOL. To find the edges corresponding to the bases
108
and the tips
106
, optimally projection is performed along a direction of a line tangent to the edges of the bases
108
and the tips
106
. Accordingly, the window is positioned such that its projection axis, p, is as close as possible substantially parallel to a line tangent to the lead bases
108
and the lead tips
106
and normal to a lead axis, T-T′, of the leads
102
. The degree p can be of offset from parallel depends upon each application and varies widely as is known in the art. The length, l, extends across each end of the leads
102
. Alternatively, two windows
112
and
114
each enclosing one end of each lead can locate each base
108
and tip
106
, where p of the two windows is also substantially parallel to a line tangent to the bases
108
and tips
106
and substantially normal to T-T′.
A problem with these methods is the lack of integrity of the generated edge information. Other structures in the image generate extraneous edges, such as feature
302
on the device body
300
and the silhouette of the device body
304
, not drawn to scale, illustrated in gullwing-leaded device of FIG.
3
. The extraneous edges confuse identification of the edges of the lead base
308
and lead tip
306
.
A further drawback of this method is that back-lit images do not have edges corresponding to both the lead base
308
and the lead tip
306
.
FIG. 4B
illustrates back-lit imagery, where the light
450
originates from a light source
458
located behind the object
452
and is directed toward the imaging device
456
so that the object
452
and the leads
454
appear as a silhouette.
Another drawback of this method is that it is not easily extended to deal with varying lead lengths.
Alternatively, the length of leads is determined by binarizing the image of the leaded device. Binarizing an image is a technique where a threshold is chosen to segment the image into foreground objects and the background. Typically, one intensity, such as white, denotes the leads, and the other intensity, such as black, denotes the image background. Once binarized, the length of the white object in the image is determined easily using known methods, such as a connected component analysis.
One of the short falls of the binarization technique is the inability of a single threshold to segment the entire lead from the background. Typically, the leads have specularly reflecting surfaces that frustrate identifying a threshold within a front-lit image of a leaded device.
FIG. 4A
illustrates a front-lit system, where a light source
408
directs light
410
towards a bottom of a leaded device
402
and leads
404
, and the light reflects off the leaded device
402
and the leads
404
back to the imaging device
406
which collects the light. The metal leads
404
specularly reflect the light
410
of the front-lit system. Further, the shape of the leads
404
causes reflections in some portions of the leads
404
to be stronger than other reflections. An example of the varying intensities imaged from a portion of a gullwing-leaded device is illustrated in FIG.
3
.
Back-lit images of the leaded devices do not exhibit specular reflections because only the silhouette of the device is imaged. In a back-lit image, the leads and the device body have substantially the same grey scale value, and, therefore, no threshold exists that segments the entire lead relative to the body and background. Consequently, the base of the leads cannot be identified in the image. Thus, the binarization method is not an optimal solution.
In addition to leads, other parallel objects that are in close proximity to each other often frustrate prior methods for measuring length of the parallel objects.
SUMMARY
Methods and apparatuses are disclosed for measuring an extent of a group of objects within a digital image by comparing a reference signature, representative of at least one relationship of the objects to one another, against instances of a measured signature representing various positions within the image. The position(s) where the signature(s) vary by more than a predetermined comparison criteria are used to calculate the extent of the group of objects. More generally, the comparison criterion indicates when the measured signature no longer represen
Bachelder Ivan A.
Marrion Jr. Cyril C.
Montillo Albert A.
Calabressi Tracy
Chawan Sheela
Cognex Corporation
Johnson Timothy M.
Weinzimmer Russ
LandOfFree
Methods and apparatuses for measuring an extent of a group... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatuses for measuring an extent of a group..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatuses for measuring an extent of a group... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3079666