Image analysis – Pattern recognition – Classification
Reexamination Certificate
1998-11-13
2003-09-23
Patel, Jayanti K. (Department: 2625)
Image analysis
Pattern recognition
Classification
C348S131000
Reexamination Certificate
active
06625318
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates generally to imaging. More specifically, the invention relates to detecting defective pixels in an image sensor.
2. Description of the Related Art
Imaging devices such as digital cameras and scanners may have as one component, an image sensor which is manufactured as a CCD (Charge Coupled Device), CID (Charge Injection Device) or CMOS (Complementary Metal-Oxide Semiconductor) device. The image sensor is composed of an array of “sense” or pixel locations which captures energy from a illuminant, often converting this energy into a concrete measure such as an intensity value. In most cases, imaging sensors will have a certain number of pixel locations which are “defective” due to fabrication or manufacturing errors. It is extremely difficult, if not impossible, to guarantee during such fabrication/manufacturing that none of the pixels in the sensor will be defective. A “defective” pixel of a sensor is one which when exposed to an illuminant will produce a different intensity value or response than that of a “fully functional” pixel when exposed to that same illuminant. In other words, the defective pixel is abnormally sensitive/unsensitive to light than a fully functional pixel. Such defects if not detected and then compensated for, may cause the captured image to be of less visually perceived quality and if prominent can detract the attention of the viewer towards the defective pixel(s).
Defects in pixel locations can be split into three categories-Stuck High, Stuck Low and Abnormal Response. A Stuck High defective pixel is one which always responds to the lighting condition by producing a high intensity value. For instance, if the intensity of pixels ranges from 0 to a high of 255, a Stuck High pixel would always respond to lighting with a value of 255, even if actual measured intensity for that location of the scene would be 25, for example, if captured by a functional pixel. A Stuck Low defective pixel is one which always responds to the lighting condition by producing a low intensity value. A Stuck Low pixel may respond with a value of 5 even though a functional pixel would show the intensity value to be 200, 100 etc. A pixel with an Abnormal Response defect has no absolute, but rather a relative variance from a functional pixel. For instance such a pixel would inaccurately respond by a particular percentage, such that, for instance, were a functional pixel would read a value X, the Abnormal Response defective pixel would respond with a value 0.25*X. The Abnormal Response is thus proportional or relative to the intensity being captured, rather than an absolute high or low. Pixels exhibiting any of these types of defects should, desirably, be corrected or compensated for.
The first step in any such compensation is the detection of which pixels are in fact “defective”. Conventionally, such detection is performed by identifying the defective pixel locations in a controlled environment, such as during quality control for the sensor as a whole, after the sensor is fabricated. The identified locations are recorded and then transferred to some non-volatile memory on the device in which the sensor is used such as on a digital camera. In modern “mega-pixel” image sensors, where the total size of the sensors have on the order of 1000 by 1000 pixels, many pixels may be defective. The extra memory needed to store the defective pixel locations adds to the total cost/time-to-manufacture of the device and also requires actual data transfer during the process of assembling/integrating the sensor into the device. The defective pixel locations must be separately stored prior to device assembly into a fixed memory such as a hard disk. Once the defective locations are stored, signal processing techniques post image capture may be used to correct the defective pixels. A more arbitrary way of correction image defects, which has also been utilized, is to not detect defective pixels, but treat the unknown defects as noise and apply an image-by-image noise removal technique to the entire sensor output (image). While avoiding memory cost and data transfer during assembly, such techniques have the disadvantage of being computationally expensive to implement and of potentially reducing the sharpness of the image, which is a key to visual appearance.
For these reasons, there is a need for a method to detect and compensate for defective pixel locations without adding to the time/cost of manufacture of the device and without sacrificing image quality or adding to the computation requirements during image processing on the device in which the sensor is to be employed.
SUMMARY
What is disclosed is a method comprising performing an observation on a sensor having a plurality of pixels, for each of the pixels that are unclassified, determining a score according to the observation, if the score for the each pixel satisfies a stopping condition, classifying the each pixel as being one of either defective or functional, and repeating the steps of performing, determining and classifying for any the pixels remaining unclassified after determining the score.
REFERENCES:
patent: 4598420 (1986-07-01), Harvey
patent: 4817166 (1989-03-01), Gonzalez et al.
patent: 5007734 (1991-04-01), Wilms
patent: 5717781 (1998-02-01), Ebel et al.
International Search Report, PCT/US99/26204, Nov. 4, 1999, 5 pages.
Acharya Tinku
Tan Yap-Peng
Azarian Seyed
Patel Jayanti K.
LandOfFree
Robust sequential approach in detecting defective pixels... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Robust sequential approach in detecting defective pixels..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Robust sequential approach in detecting defective pixels... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3112515