IRJET- 3D Object Recognition of Car Image Detection
WFIRST Poster Small File Size
1. Development of Image Analysis Routines for WFIRST Detector Characterization
Davis Unruh1
, Jonathan Mah2
, Laddawan Miko2
Bowdoin College1
, NASA Goddard Space Flight Center2
Abstract
In order to compare the performance of IR imaging detectors for astronomy applications, the
image data needs to be analyzed to produce certain figures of merit. Most notable are dark
current, noise, gain, and quantum efficiency. Depending on the operating conditions such as
temperature and voltage bias, these can vary for a particular detector. It is important to be
able to accurately represent these statistics in order to properly compare detectors. This pro-
ject examines the design and development of image analysis routines to characterize different
properties of IR detectors. Over the course of the project, these routines were developed in
the form of a graphical interface to measure these different properties for data read out of a
generic square IR detector. The figures of merit generated using this interface will be com-
pared across detectors and varying operating conditions, and used in the process of develop-
ing and improving the detectors for eventual use in the WFIRST telescope.
Acknowledgements
I would like to thank the Maine Space Grant Consortium for providing the
groundwork for the internship opportunity.
I would also like to thank Jon Mah for his help and guidance throughout the
duration of the project, as well as Laddawan Miko for her support and the op-
portunity to work with the WFIRST project.
Methods
The GUI was designed to read in sequences of files, excluding frames or
bounding the detector area as needed. This data can be masked using a
Gaussian fit, and the statistical method for choosing data points of interest
can be varied. The figures of merit that can be generated are:
Various methods have been implemented for generating these figures,
depending on the format of presentation the user wants.
Introduction and Background
WFIRST, the Wide-Field InfraRed Survey Telescope, is a space mission
designed to answer fundamental questions in the areas of dark energy,
exoplanets, and infrared astrophysics.
Essential to the operation of WFIRST is a set of large format focal plane
arrays, HgCdTe CMOS detectors, which have a high quantum efficiency
in the SWIR spectrum. These detectors are tuned to have very low
dark current, such that you can observe very faint stellar objects.
This project is part of the process of developing these detectors. It is
designed to streamline the process of applying statistical methods to
detector image data, by combining separate routines into a single GUI.
The goal is not only to compress the time scale of running image analy-
sis, but to also provide a consistent platform through which to process
image data and run analysis routines.
This ensures that data is properly displayed for comparison purposes.
Conclusion
The project was a success. The GUI does its job well: figures of merit are
now easier and faster to generate, and more standardized in how they
are generated relative to each other.
The process of altering background variables for visualization purposes
is more consistent, changes and adjustments will affect all figures
equally without any alterations necessary to the code used to generate
any figures. Combined with the variety of methods coded into the GUI
for generating figures, the GUI enhances flexibility and speed in visualiz-
ing data.
This project will be used for future detector image analysis, both for the
WFIRST program as well as other detectors. It is a step towards mini-
malizing the process of generating figures of merit, allowing focus on
the process of generating data, and analyzing results.
DETECTOR
HEATERS &
TEMP SENSORS
LN2 TANK
TEMPERATURE
CONTROLLERS
COPPER STRAPS
DEWARINTERFACE
BIAS
CLOCK
VIDEO
READOUT
ELECTRONICS
A—D
BIAS & CLOCK
GENERATOR
PC
FRAME
GRABBER
GUI
CURRENT
SOURCE
DEWAR
Detector Data Collection Set-up
Dark Current
Gain
Persistence
Noise
Signal Histograms
ROIC
FANOUTPCB
LED