Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
1999-11-24
2003-07-22
Chang, Jon (Department: 2623)
Image analysis
Applications
Manufacturing or product inspection
C382S172000, C358S466000
Reexamination Certificate
active
06597805
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a visual inspecting method for electronic device, a visual inspecting apparatus for electronic device, and a record medium for recording a program which causes a computer to perform the visual inspecting method, and in particular, to those which inspects defects on a surface of a package of an electronic device.
2. Description of the Prior Art
A visual inspecting method/apparatus for automatically inspecting a defect such as a small hole (hereinafter, referred to as “void”) that takes place on the top surface of the package of an electronic device that has been fabricated is known.
In a first prior art of such a visual inspecting method disclosed as JPA 5-280958, a photographed image of the top surface of an inspection object is divided into a plurality of unit regions. With the average value of gradation levels of each unit region, the image is binarized. Whether a defect exists is detected on the basis of the shape of a region with low reflectance in the photographed image.
FIG. 1
is a block diagram showing the structure of a defect inspecting apparatus according to the first prior art.
In
FIG. 1
, a scanning beam is radiated from laser light source
20
to the top surface of inspection object
10
. The reflected light is guided to light receiving units
28
. Light receiving units
28
output image signals a
1
and a
2
, respectively. The image signals a
1
and a
2
are supplied to adding unit
32
, A/D converting unit
34
, brightness converting unit
36
, and differentiating filter
38
in the order. Differentiating filter
38
outputs signal a
5
in which the contour of a defect is emphasized. Signal a
5
is supplied to defect address detecting unit
40
. Defect address detecting unit
40
obtains a defect address signal Ad from signal a
5
. Signal a
5
is also supplied to binarizing unit
42
. Binarizing unit
42
binarizes signal a
5
. The binarized signal is supplied to defect region extracting unit
48
. Defect region extracting unit
48
extracts a defect region on the basis of the binarized signal and defect address Ad and outputs the extracted defect region as defect image signal A
2
.
In the defect determining method of the first prior art, the address of a defect is obtained on the basis of the differentiated image of the inspection object
10
. Binarizing unit
42
performs the following calculation.
TH
3
=
TH
1
−k
(
TH
1
−
TH
2
)
where TH
1
is a first threshold value that is obtained from the distribution of gradation levels of the entire differentiated image; TH
2
is a second threshold value obtained from the average value of the gradation levels of adjacent regions of a considered pixel; and TH
3
is a third threshold value. In addition, binarizing unit
42
binarizes the gradation level of each pixel by using threshold TH
3
.
The defect region extracting unit
48
determines that a defect on the top surface of the inspection object
10
is present at a pixel position whose gradation level is lower than threshold value TH
3
.
FIG. 2
is a schematic diagram showing each region of the inspecting apparatus shown in FIG.
1
.
FIG. 2
shows an example of inspection object
10
shown in
FIG. 1
which is a concrete electronic device
2
. Referring to
FIG. 2
, a photographed image that is output from light receiving unit
28
contains a package
2
a
and a part of terminals
2
b.
A region that contains only the package
2
a
is designated as an inspection objective region Rt. The photographed images of package
2
a
contain an image of void B that is a defect on package
2
a,
images of marking characters T marked on package
2
a,
and an image of fluctuating portion P formed on the top surface of package
2
a
or formed due to reflected light.
FIG. 3
is a graph showing characteristic curves of gradation levels Lc, binarization levels Ls, versus pixel coordinate positions of the defect inspecting apparatus shown in FIG.
1
.
FIG. 3
shows levels taken along line X—X of
FIG. 2. A
sharp concave portion on the left of the curve of the gradation levels Lc of the photographed image represents a void B. On the other hand, a broad concave portion on the right of the curve of the gradation levels Lc represents a fluctuating portion P. Three protrusion portions in the middle of the curve of the gradation levels Lc represent marking characters T.
At the void B, the peak width is narrow and the curve of the gradation levels Lc sharply varies. At the fluctuating portion P, the peak width is wide than that. of the void B and the curve of the gradation levels Lc gradually varies. At each of the marking characters T, although the peak width is narrow, the curve of the gradation levels Lc is higher than that of the package region.
In
FIG. 3
, the “1” level region where gradation level Lc is higher than binarization levels Ls is determined as a normal region (no-void region), whereas the “0” level where gradation levels Lc is lower than binarization levels Ls is determined as a void region.
As is apparent from
FIG. 3
, at fluctuating portion P, although the curve of gradation levels Lc slightly lowers in a wide range, the curve of binarization levels Ls obtained as the average value of curve of the gradation levels also gradually lowers.
Next, a second prior art of such a defect inspecting method will be described. The second prior art is simpler than the first prior art. In the second prior art, all the photographed image of the top surface of a package is binarized with a single predetermined binarization level. When the area of the “0” level region of the digitized image (namely, the area of a region whose reflectance is small) is higher than a predetermined value, the region is determined as a void region.
FIG. 4
is a graph showing characteristic curves of gradation levels Lc, binarization levels Ls versus pixel coordinate positions of a defect inspecting apparatus according to the second prior art. Similarly to
FIG. 3
,
FIG. 4
shows curves of the photographed image of package
2
a
as shown in FIG.
2
. to As is apparent from
FIG. 4
, the curve of binarization levels Ls is constant in all the range.
In the conventional defect inspecting method/apparatus, it was determined whether or not a defect such as a void is detected on the top surface of a package of an electronic device in a manner as explained above.
However, the conventional defect inspecting methods/apparatuses have the following disadvantages.
When electronic device
2
is used as an inspection object, the intensity of light radiated from light radiating source
1
to the top surface of the package
2
a
may vary in dependence on the direction and location of light radiating source
1
. In addition, the reflectance of the top surface of package
2
a
may vary position by position because of fluctuation of the ingredients and surface condition of the resin of electronic device
2
and a stain adhered from a die or the like. In these cases, the gradation levels Lc of image signals al and a
2
in a part of of the package
2
a
may differ from those of other parts as shown in
FIGS. 3 and 4
. Thus, fluctuating portion P tends to take place.
When the gradation level of fluctuating portion P is almost the same as that of void B, after the fluctuating portion P is binarized, it may be determined as the “0” level region. Thus, even if package
2
a
does not have a void, it is often mistakenly detected.
In the defect inspecting apparatus according to the first prior art, as shown in
FIG. 3
, the average value of gradation levels Lc of each unit region is used as a binarization level Ls. However, in this method, in the vicinity of a marking character T shown in
FIG. 3
, since gradation level Lc of the marking character T is high, the binarization level Ls becomes high. Thus, in the vicinity of a marking character T, a void may be mistakenly detected.
On the other hand, in the defect inspecting apparatus according to the second prior art, as shown in
FIG. 4
, the binarization process is performed for al
Chang Jon
Katten Muchin Zavis & Rosenman
LaRose Colin
NEC Corporation
LandOfFree
VISUAL INSPECTION METHOD FOR ELECTRONIC DEVICE, VISUAL... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with VISUAL INSPECTION METHOD FOR ELECTRONIC DEVICE, VISUAL..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and VISUAL INSPECTION METHOD FOR ELECTRONIC DEVICE, VISUAL... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3063892