Increasingly, the signature management community is demanding modeling tools for a variety of purposes from real-time simulations to complex modeling tasks. RenderView is one of the tools which has been developed and continues to evolve in response to this demand.
The focus of RenderView development has been physics based modeling of high complexity both geometrically and with respect to surface optical properties. RenderView incorporates full bi-directional reflectance distribution function (BRDF) models and measured and calibrated global illumination maps. With these tools comes the capability to evaluate with a very high level of fidelity the impact of vehicle geometric and surface properties on its visible and thermal signature.
A description of RenderView will be presented in terms of its focus on high fidelity models of vehicles and materials. A number of examples will be presented that show how the fidelity of the BRDF impacts the signature via the rendering model.
As part of the survivability engineering process it is necessary
to accurately model and visualize the vehicle signatures in multi-
or hyperspectral bands of interest. The signature at a given wavelength is a function of the surface optical properties, reflection of the background and, in the thermal region, the
emission of thermal radiation. Currently, it is difficult to
obtain and utilize background models that are of sufficient
fidelity when compared with the vehicle models. In addition, the
background models create an additional layer of uncertainty in
estimating the vehicles signature. Therefore, to meet exacting rendering requirements we have developed RenderView, which incorporates the full bidirectional reflectance distribution function (BRDF). Instead of using a modeled background we have incorporated a measured calibrated background panoramic image to provide the high fidelity background interaction. Uncertainty in the background signature is reduced to the error in the measurement which is considerably smaller than the uncertainty inherent in a modeled background. RenderView utilizes a number of different descriptions of
the BRDF, including the Sandford-Robertson. In addition, it
provides complete conservation of energy with off axis sampling. A description of RenderView will be presented along with a methodology developed for collecting background panoramics. Examples of the RenderView output and the background panoramics will be presented along with our approach to handling the solar irradiance problem.
A new iterative algorithm (EMLS) via the expectation maximization method is derived for extrapolating a non- negative object function from noisy, diffraction blurred image data. The algorithm has the following desirable attributes; fast convergence is attained for high frequency object components, is less sensitive to constraint parameters, and will accommodate randomly missing data. Speed and convergence results are presented. Field test imagery was obtained with a passive millimeter wave imaging sensor having a 30.5 cm aperture. The algorithm was implemented and tested in near real time using field test imagery. Theoretical results and experimental results using the field test imagery will be compared using an effective aperture measure of resolution increase. The effective aperture measure, based on examination of the edge-spread function, will be detailed.
The spatial resolution of under sampled or diffraction limited images can be improved through micro scanning and super-resolution technologies. The objective of this Air Force Phase Ii Small Business Innovative Research was to develop and demonstrate real-time or near real-time micro scanning and super-resolution algorithms using passive millimeter wave imagery. A new super-resolution algorithm based on expectation-maximization was developed which is insensitive to missing data, incorporates both positivity and smoothness constraints, and rapidly converges in 15 to 20 iterations. Analysis using measured data shows that the practical resolution gain that can be expected using this algorithm is less than a facto of two. A new micro scanning algorithm was developed and demonstrated that can reliably detect less than one fifth of an IFOV displacement using field test data. The iteration of the super-resolution and microscanning algorithms was demonstrated and resolution gains of four to six times can be achieved if the image is under sampled by a factor of two or three. Consequently, it makes sense to use a wide under sampled FOV sensor in which high spatial resolution can be obtained as desired using micro scanning and super-resolution techniques.