Autonomous focal plane calibration by an intelligent radial basis function network
Conference Paper
Overview
Identity
Additional Document Info
View All
Overview
abstract
In this paper, a novel focal plane calibration algorithm is presented using a multi-resolution learning algorithm, that can be implemented online. For global learning of the distortion map, the use of a radial basis function network is advocated whereas a local approximation method known as moving least squares is used for microscopic learning of the distortion map. The algorithm proposed in the paper is inherently sequential in nature and therefore can be used recursively in real-time to update the calibration coefficients. The network architecture itself is adapted in contrast to adjusting weights in a fixed architecture network. The algorithm is validated by introducing different kind of unknown distortions in the simulated star images.