Detector noise analysis (DINO)

DINO is designed to identify periodic noise features induced by the electronics during detector readout. DINO is a general purpose tool and should work with any format CCD with minimal changes.

Method

1. If given a list of bad columns these columns are masked to zero.
2. If a median width is specified and greater than zero.
   • Apply a median filter of the specified width on each column individually.
   • Trim the first and last median width / 2 rows to avoid median filter end effects.
   • If the cosmic ray rejection sigma is greater than zero then mask pixels above sigma standard deviations (computed in trailing 15 columns) to zero.
3. Pad each row with pst/sst pixels to represent the parallel shift interval, where pst and sst are the parallel and serial shift times respectively.
4. Deconstruct the two dimensional image into a one dimensional time series vector in which each element represents an equal time interval.
5. Trim the time series vector to a power of two for faster FFT processing.
6. Compute the FFT of the time series vector.
7. Plot the power spectrum versus frequency. Power is computed as the square of the magnitude of the complex transform.

Automatic processing

DINO_LOAD is a helper tool that loads a new data set into an already running DINO. When DINO_LOAD is used in conjunction with a tool (such as LATEST) that can execute a command every time the ACS preflight database is updated, DINO processing becomes completely automated. And for each amp of an image it produces output like figure 1.

Figure 1.