Adaptive Airborne Infrared Non-Uniformity Correction

HNC has developed a VLSI image processing system and associatetd neural network software capable of rapidly performing non-uniformity correction for an infrared focal plane array sensor using the adaptive neural network algorithm proposed by Dean Scribner and his colleagues at NRL. HNC has implemented this algorithm in the laboratory using HNC's Vision Processor (ViP) chip set was developed by HNC and ARPA to perform high speed image and neural network processing. The ViP consists of two VLSI gate arrays that can perform fundamental processing operations such as large kernel convolution, subtraction of two images, multiplication of two images and look-up table transformation of an image. A daughterboard with the ViP is built and is available from HNC as a commercial product. HNC proposes to extend our preliminary work with NRL algorithm, analyze its restrictions and precision requirements, investigate improvements, compare its performance to other possible algorithms, and since we have existing hardware that can perform non-uniformity correction in close to real-time, demonstrate and test that hardware using MWIR or LWIR imagery.