Image analysis – Image enhancement or restoration – Artifact removal or suppression
Reexamination Certificate
2001-09-21
2003-03-11
Lee, Thomas D. (Department: 2624)
Image analysis
Image enhancement or restoration
Artifact removal or suppression
C382S199000
Reexamination Certificate
active
06532310
ABSTRACT:
FIELD OF THE INVENTION
The invention is directed to improving the output obtained from an imaging system and, in particular, to a technique which processes a digitized image of an object to remove noise.
BACKGROUND OF THE INVENTION
Imaging systems are used in such fields as microelectronics, medicine, biology, genetic engineering, mapping and even astronomy. The imaging device can be a suitable type of microscope or, in the case of astronomy, a telescope. The demand for image accuracy is high and, therefore, the influence of noise in a signal derived from an imaged object must be minimized.
For reasons of convenience and efficiency, the invention will be described in the microelectronics environment, although another environment could also have been chosen. During the manufacture of very large scale integration (VLSI) semiconductor devices, measurements are made at several stages of the manufacturing process to determine whether particular features on the object are within specified design tolerances. If not, then suitable corrective action is taken quickly.
As is well known, such a manufacturing process produces a wafer which is divided into dies. Each die has a large number of electronic components. These components are defined by what can generally be termed “features” in the sense that a feature is detectable by a microscope as a foreground element distinguishable from a background, or vice versa, and having a dimension such as width. To measure that width the edges of the feature must be located accurately. “Edge” is a term used to signify detectable discontinuities in a signal obtained by imaging the feature (in any environment, not only microelectronics). The goal of edge detection is to accurately locate the transitions despite the influence of blurring and the presence of noise.
As technology has succeeded to increase the component density per die, the feature dimensions have shrunk to significantly below a micrometer. Consequently, the measurement equipment must measure submicrometer dimensions with lower allowable error tolerances.
Automated systems have been developed for making these measurements to replace manual systems in order to obtain higher process yields, to reduce exposure of the wafers to contamination and to provide a higher throughput. One example of an automated system is disclosed in U.S. Pat. No. 4,938,600. As shown in
FIG. 1
which is taken from that patent, an image of a feature is recorded through a microscope and the recorded image is then processed electronically to obtain the required measurements. One such automated system is the Model IVS-120 metrology system manufactured by Schlumberger Verification Systems of Concord, Mass., a division of Schlumberger ATE Products. The major elements of the system, including a wafer handler, an optical system and a computer system, are mounted in a cabinet (not shown).
The wafer handler includes a cassette wafer holder
12
which contains wafers to be measured, a prealigner
14
, a wafer transport pick mechanism (not shown) for moving the wafers and a measurement stage
18
which holds the wafers during the actual measurement operation. During operation, the wafer transport pick mechanism removes a wafer
16
from cassette
12
and places it on prealigner
14
. Prealigner
14
then rotates wafer
16
to a predetermined orientation by sensing a mark, a flat spot or notched edge on wafer
16
, after which the wafer transport pick mechanism transfers wafer
16
from prealigner
14
to measurement stage
18
and positions wafer
16
in a horizontal orientation. Stage
18
is movable in three dimensions for precisely positioning wafer
16
relative to the optical system for performing the actual measurement.
The optical system includes microscope
20
and video camera
22
positioned above the measurement stage
18
and wafer
16
. Microscope
20
typically has a turret carrying several objective lenses providing a desired range of magnification and is mounted so that microscope
20
and camera
22
have a vertical optical axis which is perpendicular to the wafer surface.
A feature to be measured on wafer
16
is located with microscope
20
in a well known manner by moving stage
18
until the feature is in the field of view of the objective lens. The optical system is focused, and a focused image of the feature is digitized and recorded by the camera
22
. The image is then stored or “frozen”.
The system is controlled by a computer
30
. Coupled to the computer
30
are a monitor
32
for display of the image recorded by the camera
22
and text, and a keyboard
36
(which constitute an input terminal for entering operator commands) and a disk drive
38
for storing system software and data.
Image processor
28
uses software algorithms to locate the edges of the selected feature and make a measurement. Computer
30
then displays the measurement data on screen
32
, prints a hard copy or transfers the data directly to a host computer (not shown) for centralized data analysis. Once the process is complete, wafer
16
is returned to cassette
12
by the wafer handler.
The just-described system performs its task of edge detection very well. Image processor
28
determines where a discontinuity occurs in the gray level of the digitized image. Such a discontinuity can occur for any one of many well known reasons to create an edge of a feature. For example, an edge can occur where two materials meet which have different gray levels, or due to topology of the imaged surface. However, as is well known, the digitized image is subject to spurious noise from various sources. For example, variations in the gray level due to noise can be caused by surface imperfections on the die, such as spots and cracks. This noise in the imaged signal can have a significant distorting influence on the accuracy with which the edge is detected, particularly with the ever increasing precision which such automated measurement systems must provide. (Of course, in environments other than microelectronics, there are analogous causes of noise.)
Certain approaches are known which aim to eliminate the noise created by these imperfections and thereby improve the signal-to-noise ratio (S/N). For example, smoothing filters are commonly used for noise reduction. However, as explained in
Digital Image Processing
by Gonzales and Woods, Addison-Wesley Publishing Co. 1993 at page 191, a smoothing filter blurs edges because it relies on neighborhood averaging which averages all the pixels in an area of selected size around a pixel of interest. Such a blurring of the edge cannot be tolerated in a measurement system which must locate the edge precisely. For such an application, the authors recommend an alternative approach which uses median filters. This approach replaces the gray level of each pixel by the median of the gray levels in a neighborhood of that pixel, instead of by the average. This method is particularly effective to preserve edge sharpness when the noise pattern includes strong, spike-like components. However, even median filtering is not satisfactory for the type of precision measurements discussed above because when the filter parameters are set to provide filtering, the edge gets modified, and when the parameters are set to preserve the edge, the filtering effect is reduced or even eliminated.
SUMMARY OF THE INVENTION
One object of the present invention is to remove the influence of noise in a signal obtained with an imaging system.
A further object of the present invention is to improve the S/N ratio of the output signal received from the imaging device in such system.
Another object of the present invention is to enable improved accuracy for a high precision measurement system which uses an imaging device.
Yet another object of the present invention is to enable improved accuracy of edge detection with an imaging system despite the presence of noise in the image signal.
These and other objects are attained in accordance with one aspect of the present invention for reducing noise in a signal obtained by imaging an object with an im
Brinich Stephen
Lee Thomas D.
Schlumberger Technologies Inc.
LandOfFree
Removing noise caused by artifacts from a digital image signal does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Removing noise caused by artifacts from a digital image signal, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Removing noise caused by artifacts from a digital image signal will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3058190