Polarimetric sensors are valued for their capability to distinguish man-made objects from surrounding clutter. The
SPITFIRE (Spectral Polarimetric Imaging Test Field InstRumEnt) polarimetric camera is designed to function in
multiple bands in the Short Wave Infrared (SWIR) and Mid-Wave Infrared (MWIR) regions. SPITFIRE is a Stokes
micro-grid polarimetric system with a 4 band spectral filter wheel. The focal plane array (FPA) as well as the filter
wheel are located in a Dewar which is cooled via liquid nitrogen. By cooling the band-pass filter to the same
temperature as the FPA, self-emission noise is decreased. In this paper we discuss the design and fabrication of the
polarimetric camera (optics, Dewar, filter wheel and FPA), the data capture and processing system, initial
characterization of the camera's performance, and future plans for the camera.
Simultaneous detection of intensity and polarization at the pixel-level has many important applications in the mid-infrared
region. In this work a large-format aluminum wire grid micro polarizer array has been fabricated and tested on
silicon substrates. The arrays were made on 150mm silicon wafers using a 193nm deep-UV stepper, with each array
spanning over 1-million pixels. A unique multilayer design and a large-area nanoscale projection lithography combined
with high-aspect ratio wire-grid structures were utilized to achieve optimum extinction coefficient and transmission.
Measured extinction coefficients on test samples exceeded 30-dB, with maximum transmission around 90%. These
arrays could be designed to match the focal-plane array geometry for integration with mid-IR imagers.
Past work with polarimetry in the mid-wave infrared (MWIR) has yielded mixed results. In order to better characterize
polarimetric content in the MWIR and short-wave infrared (SWIR) atmospheric windows, we are developing focal
plane array (FPA) technology that will address shortcomings in earlier devices. In particular, our efforts are focusing on
placing micro-polarizing grids in very close proximity to the P-N junction of the detector. By placing these micropolarizers
very close to the photodetector junction, the opportunity for polarimetric cross talk between pixels is
minimized. CE's unique process for fabricating FPAs is well suited for implementing this approach. Since a
polarimetric FPA consisting of a standard FPA and micro-wire grid polarizers reduces the effective FPA format by a
factor of two in both dimensions, the ability to produce extremely large format FPAs are critical to obtain high
resolution polarimetric imagery. CE's FPA fabrication process is also highly scalable and has successfully fabricated
FPAs as large as 2k by 2k. This paper describes the progress we've made towards developing these unique polarimetric
Birefringent optical materials can be used to convert mechanical strain into fringe patterns of optical intensity which have typically been used to measure surface stains or stresses. In this paper a system will be described that uses a photoelastic transducer, linear sensor array, and neural network image processing to estimate the load torque for stationary and rotating motor shafts up to 1500 rpm. A photoelastic polymer coupling is attached to the shaft, and illuminated by polarized light. As the shaft torque varies the photoelastic plastic coupling experiences torsional strain. This results in a corresponding 2D fringe pattern when viewed through an optical polarizer. The strain that causes this observed pattern in a complex function of the applied torque applied to the shaft. A neural network is trained with the fringe patterns corresponding to calibrated load torques as measured by a laboratory strain gauge torque sensor. Experimental results show that the neural network torque estimator can accurately estimate the applied torque for both static and rotating shafts.
The results of a program to develop a low-cost, real-time approach for 100 percent inspection of fast-moving process lines for textures, paper, and steel are reported. Initial results will be described of a laser based scanning system that uses a neural network for training and the implementation of a neural network with an optical processor for real-time defect inspection.