Huy Tuong Cao (remotely), Won Kim (remotely), TVo, Danny 

We've recently come into the issue of an unusually noisy ITMX HWS with noise on the order of 50 microdiopter. It was noted in alog 44447 that hot/dead pixels may be to blame for a good fraction of ITMX HWS noise. We decided it would be worth it to enable dead/hot pixel subtraction within the HWS analysis code.

The following is a simplified "how to" on Hartmann image pixel subtraction: 

1. Cover up the HWS camera and run the tale command on the HWS computer to take 100 dark images and deposit them in the appropriate dark images directory. This "dark_images" directory may not exist yet on other computers so you may need to make it. (i.e. "take -c 1 -N 1 -s 1 -l 100 -f /home/controls/camera_folder/14011236/dark_images/ -m 1000") 
2. Write a script to generate a numpy array containing the coordinates of the dead pixels. Functioning example code is written in /home/controls/HWS_code_temporary_home/common/hws_code/gen_bad_pixel_array.py in the h1hwsmsr computer. 
   • This code takes the dark images generated in step 1, averages them, and stores coordinates of particular pixels that exceed a specified threshold within a numpy array. The suggested threshold value is 100 (this might require more fine tuning over time). 
3.  We then enable the pixel subtraction software by setting "to_fix_bad_pixels=TRUE" and specify the directory of the numpy array in State_3.py. 
   • This subtraction removes the bad pixel from the image and assigns a new value that is the average of the nearest neighbors. 

After implementing this, the noise is now approximately 5 microdiopters.