- Occulting fingers located near the aperture entrance if the image size is greater than 512 x 512 pixels or if the FUV prism is in the beam.
- Reseau marks etched onto the faceplate of the detector to aid in geometric correction (Figure 4.4).
- Blemishes (scratches on the faceplate, Figure 4.4).
- Vertical intensity variations along the right edge of the image (due to a -variation in camera scan speed).
- A faint diagonal parallel striping pattern called pattern noise.
After geometric correction your images may additionally show:
- A very faint moiré pattern ("thumbprint"), which is a variation of the noise, not the signal, caused by the geometric correction (Figure 4.5).
- Warped edges (Figure 4.4).
These features are all normal and should be expected. They can be traced either to the instrumental design and performance of the FOC or the calibration process which corrects for geometric distortion.
Figure 4.4: FOC f/96 Image of an Extended Source
Figure 4.4 is a positive rendition of an f/96 (F430W) 512 x 512 image of the reflection nebula LK-H alpha 233. The image has been fully calibrated by the FOC pipeline and shows features that are common in any well-exposed FOC image. The regular grid of reseau spots are used for geometric distortion calibration. The warped edges are produced by the pipeline during the geometric transformation. A blemish is seen above and to the right of center.
Figure 4.5: FOC 512 x 512 Image Showing Faint "Thumbprint" Pattern
or "Fringes"
Images that have been geometrically corrected often show the pattern evident in the above grayscale picture: a thumbprint pattern at low intensity levels. It is quite hard to see, appearing most clearly when the image has a low (~1 count per pixel or so) spatially flat background. The thumbprint pattern is a modulation of the local noise characteristics of the data, not of the intensities themselves. It is a by-product of the geometric correction process in which the raw (geometrically distorted) image is resampled with an interpolator that takes a weighted mean of the four nearest pixels to determine the geometrically corrected pixel value (see "Geometric Correction (GEOCORR)" on page 6-5 for details). The weightings vary smoothly with position in the image, such that at some places, a single pixel dominates the weightings (the noise of the resampled pixel is the same as that of the original data), while at other places the weightings favor all four pixels equally (the average noise is half of the noise of the individual pixels). The fringes are contours of constant weighting.
The actual pattern depends on the particular geometric correction file used, and thus depends on the format. The effect on the scientific utility of the data is minimal, unless one requires accurate values of the noise per pixel for each pixel.
