Keith Knox (Boeing LTS Inc.)
Keywords: Imaging
Abstract:
When a pixel has saturated, its value has exceeded the dynamic range of the analog-to-digital converter. All that is recorded is that the value of the individual pixel exceeded the maximum value of the A/D converter. The actual value of the pixel is lost. There is one circumstance, however, in which the pixel value can be recovered, and that is when the image has been blurred by frame transfer smear.
Frame-transfer CCD sensors make an exposure and then transfer the charge image to a readout buffer. During the time it takes to transfer the charge to the readout buffer, the light-sensitive array continues to be exposed to the image. This exposure during the motion of the charge causes a linearly-smeared component to be added to the image, which is roughly proportional to the ratio of the transfer time to the exposure time.
Ordinarily, this smear would be considered a defect that needs to be avoided, but it actually opens an opportunity to recover the values of any saturated pixels. Because the image has been smeared across the CCD array, the brightest areas of the image that saturated in the exposed image have also been imaged all across the array in regions where it is not saturated.
This paper will analyze the effects of saturation in individual pixels on the smear correction algorithm. The analysis shows that a residual streak in the readout transfer direction remains after the smear correction is performed. The value of the streak is proportional to the loss in pixel value due to the saturation. By measuring the amount of the residual streak, a correction value to the saturated pixels can be determined.
The analysis of the effects of saturated pixels on the smear correction algorithm will be given and the conclusions will be illustrated with examples from saturated images of unresolved stars.
Date of Conference: September 12-15, 2007
Track: Imaging