Television – Monitoring – testing – or measuring – Testing of camera
Reexamination Certificate
1998-10-30
2003-07-15
Garber, Wendy R. (Department: 2612)
Television
Monitoring, testing, or measuring
Testing of camera
C348S247000
Reexamination Certificate
active
06593961
ABSTRACT:
FIELD OF THE INVENTION
The present invention pertains generally to image processing, and more particularly to a method and apparatus for classifying the image quality of an optical sensor using additional image characteristic thresholds.
BACKGROUND OF THE INVENTION
A digital image is an array of pixels, each of which is digitally quantized into an intensity level. Digital images are known to require a significant amount of storage area. For example, in a black-and-white image, a digital image comprising an array of 512×512 pixels, each quantized into a gray level represented by 8 bits, requires 0.25 megabytes of storage. An increase in the sampling rate (i.e., number of pixels in the array) and/or an increase in the quantization level (i.e., the number of bits, or gray, levels, that represent the image) further increases the amount of required storage for an image. The use of color, which is typically formed by varying the intensity levels of a combination of the three primary colors, red (R), green (G), and blue (B), and collectively known in the industry as the RGB components, also increases the amount of storage required. Accordingly, due to the large amount of data generated for a single image, the use of high-resolution digital images in computer applications was generally, until recently, somewhat limited.
With the leaps in amount of processing power and data storage capability in modern computer systems over the past few years, along with the lower costs and therefore increased availability to more end-users, the use of higher-resolution (represented by a larger number of pixels), higher-quality (represented by increased number quantization levels), and even color, digital images has become more and more viable.
At the same time, leaps in modern digital telecommunication system technology, including improvements in network bandwidth and network switching hardware for both the Internet and digital cellular systems, has increased both business and personal communication over networked computers. This has increased the demand for, and use of, digital images as a part of modern remote communication.
With the increased use of digital images in computer applications and communication, increased attention has been focused on the quality of the image. Digital images may be generated in a number of ways, including by way of microdensitometers, flying spot scanners, image dissectors, vidicon cameras, and photosensitive solid-state array sensors.
Microdensitometers and flying spot scanners receive input in the form of analog transparencies such as a film negative or a photograph, and digitize them into a digital pixel file. With image dissectors and vidicon cameras, an image is focused on a photosensitive tube whose response is proportional to the incident light pattern. The image dissector employs an electromagnet which is used to deflect the entire beam past a pinhole located in the back of the dissector tube. The pinhole lets through only a small cross-section of the beam which is quantized into a single pixel. The electromagnet must move the beam for each discrete pixel desired. With the vidicon camera, the image focused on the tube surface produces a pattern of varying conductivity that matches the distribution of intensity in the optical image. The conductivity pattern is scanned by an electron beam which converts into a signal proportional to the intensity pattern of the input image. This signal is then quantized to produce the digital image.
By far the most common method of generating a digital image is via solid-state arrays. Solid-state array sensors comprise an array of discrete silicon imaging elements called “photosites”, and are generally organized into one of two geometrical arrangements. The first, known as a “line-scan sensor”, comprises a row of photosites and produces a two-dimensional image by relative motion between the scene and the detector. The second, known as an “area sensor” is a two-dimensional matrix of photosites and captures the image in a similar manner as a vidicon camera.
In the manufacture of digital image sensors, the quality of the sensor is often categorized into different levels of classifications that are based on the image quality. For example, in solid-state array sensors, it is not uncommon to incur a defect in one or more of the photosites composing it, and a determination must be made as to whether the sensor should be kept or discarded. The classifications reflect the quality of the image produced by the array sensor, and are used to indicate the appropriate applications for which a particular array sensor is suited.
FIG. 1
is a two-dimensional pixel diagram illustrating a blank image
10
that contains several common types of defects. Defective pixels are darkened in this illustration. One type of defect is known as “point” defect. A point defect is a defect in a single pixel, such as pixels
12
and
20
, with no defects in all of its N-nearest neighbor pixels. In a two-dimensional array, a pixel p at coordinates (x, y) has four horizontal and vertical neighbors located a unit distance from p at (x+1, y), (x−1, y), (x, y+1), and (x, y−1), and four diagonal neighbors located a unit distance from p at (x+1, y+1), (x+1, y−1), (x−1, y+1), and (x−1, y−1). As will be appreciated by those skilled in the art, pixels located along the borders of the image do not have a full set of neighbors since the position of some of their calculated neighbors reside outside the image.
A common method of specifying image quality is based on setting a single lower (dark) and/or upper (bright) threshold limit relative to the mean for dark and/or bright pixel intensity.
FIG. 2
is a graph illustrating a prior art single-threshold defective pixel system. In this system, pixels with intensities below the dark and/or above the bright threshold limit are flagged as defects. FIG.
3
(
a
) is a perspective view, and FIG.
3
(
b
) is a side view, of the pixel image of
FIG. 1
, where the quantized intensity level is plotted along the z-axis.
Image quality is categorized into classifications based on the size of the largest allowed cluster and the number of single pixel and cluster defects. Typically a point defect, illustrated at
12
and
20
in
FIGS. 1
,
3
A, and
3
, is imperceptible by the naked eye. Accordingly, a sensor with one or a few point pixels, while it does contain defects, may actually be of high enough quality for many applications and therefore could be used for some applications.
Another type of defect is known as a “cluster” defect. A cluster defect is a grouping of physically adjacent point defects, such as those collective grouping of pixels shown at
14
.
A disadvantage of the single-threshold method is that it may fail to detect local non-uniformity or “area defects” that are easily observed by the human eye. For example, a portion of the collective pixels at
16
and
18
are above the dark threshold level, so a cluster defect is not detected using the single-threshold method. However, a portion of the neighboring pixels are sufficiently dim that when viewed collectively by the human eye, an area defect is easily observable. Setting a tighter dark (bright) threshold level aids in the area defect detection and cluster description, but this results in a larger number of defects and a definition of a single pixel defect that is not perceived as a defect under normal viewing conditions.
One method for basing classifications on human perception is by detecting local non-uniformity by computing near-neighbor density averages over small subsets of the total array and setting limits on the local average. However, computing local averages requires the collection, and manipulation, of large amounts of data, which results in lengthy testing time and high testing costs.
Accordingly, a need exists for an improved,image classification method that is based on human perception without the associated high test costs.
SUMMARY OF THE INVENTION
The present invention is a novel method an
Agilent Technologie,s Inc.
Garber Wendy R.
Whipkey Jason
LandOfFree
Test efficient method of classifying image quality of an... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Test efficient method of classifying image quality of an..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Test efficient method of classifying image quality of an... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3107592