ISOCAM Interactive Analysis User's Manual Version 5.0 - ISO - ESA

134 CHAPTER 14. IMAGE ANALYSIS AND DISPLAY
number. If the IMAGE has had no dark correction or is badly dark corrected a regular
saw-tooth pattern will be easily discernible in the lines away from any source signal. If the
CUBE has good dark correction then the plot should just reveal background noise.
• Stabilisation: CAM’s CCD detectors can be slow to stabilise when the intensity of the
incident light changes. This is most evident when a strong source comes in to a pixel’s
FOV. In x3d click on the button temporal cut andthenonanybrightpixelintheimage
window. If the pixel contains a glitch, then you may see that it takes several read-outs
to recover. If the pixel contains a source then it is very likely that you will see a slow
response or transient behaviour in the pixel to the source detection.
If the PDS has been calibrated, without transient fitting, i.e. with s90 or m90, then
click on the button mask to look at pixels flagged in the MASK 5 . The detected unstable
pixels and dead pixels (column 24 in the LW detector) will now be marked in the plot
window. If calibration has been performed well then only those pixels which exhibit
transient behaviour will be masked.
Other methods of stabilisation which involve transient fitting, i.e. fs, inv and vision, may
flag pixels in the MASK that are to be discarded in future analysis and modify other
unstable pixels in the CUBE. The flagged or masked pixels can of course still be seen by
clicking on mask and the temporal behaviour of the modified pixels of the CUBE can be
compared with that of the unmodified CUBE.
• Glitches: Looking at any single 2D CAM image it may be easy to confuse a glitch with a
source. However, if you look at the temporal behaviour of a pixel in the IMAGEs accrued
by CAM in a STATE (i.e. DATA in an SCD), or in the CUBE from a PDS, then glitches
are usually easily distinguishable from sources. A CAM observation may be made of many
STATEs, each STATE containing many IMAGEs of the same sky position. In principle,
when CAM is pointing at a source then source signal will be present in all the IMAGEs
of that particular STATE, but most glitches are of such a transient nature that their
duration will be quite short, though intense, appearing in one or possibly a few IMAGEs. 6
Therefore in most cases, sources will appear in temporal space as relatively low-intensity
long duration signal and glitches will appear as spiky intense events.
Click on temporal cut. Scroll through the images of the CUBE until you find an IMAGE
containing a pixel of high intensity. Click on the pixel and a plot of its history will appear
in the plot window. The shape of the plot should reveal whether the selected pixel of the
current IMAGE was hit by a cosmic ray or is a source detection. A well deglitched CUBE
will have few or no spiky pixels.
Glitches can be manually removed by clicking on the button glitch and then clicking on a
pixel in the image with the middle mouse button. The selected pixel will be masked and
the MASK will be automatically updated (note that the MASK should not be supplied
to x3d as an expression if the update is to be saved). See Section 20.2.2 for alternative
manual deglitching methods within CIA. Note that there are two other manual deglitchers,
man mask and sl viewcube, inthecontrib directory of CIA.
5 If !mask=1 (complex) then all instances of non-zero mask are indicated
6 All of this depends of course on such factors as the integration time, the number of frames per state, the
response of detector to the source and the intensity of the glitch.