hifurrows

The original version of the program works by finding the correlation between the columns in an input image. With a Channel 0 image, the first five columns are examined; in a Channel 1 image, the last five columns are examined (Channel 2 images are not supported).

If the statistical results suggest a furrow is detected (i.e. the correlation between two columns is below a certain threshold), then the affected column is replaced with Null pixels. If an input image does not have a furrow, none of its pixels will be altered. The label of all output files will include a new group called "Correlations." This group lists the correlations between the relevant columns. For example, "Column3to2 0.987..." means there is a strong positive correlation between columns 2 and 3 (and very unlikely to have a furrow in the area).

For the newer version, the processing has been taken from the Version 1.42 of the original HiCal pipeline perl script and processes images with summing modes 2 and 4 only.

Furrows are checked only in the columns which are based on Channel and the Summing mode. Each specified column will have a Threshold DN based on the Summing mode and CcdId. Then the starting sample and number of samples in the specified columns are determined based on the Summing mode and Channel. The DN value of pixels in these samples are checked to be in the 0 and respective Threshold range. If not, they are determined to be a furrow and set to NULL. If a furrow was found, then trimfilter and lowpass(if option is set) are run on the image in a pipeline to smooth the edges which were Nulled due to the furrows.