Adaptive non-uniformity compensation using feedforward shunting and wavelet filter

Abstract

A system and method adapted for use with a focal plane array of electromagnetic energy detectors to receive first and second frames of image data from electromagnetic energy received from at least a portion of a scene. The first frame is a focused frame and the second frame is a blurred frame. In a feed-forward path the system compares the first frame to the second frame and provides an error signal in response thereto. In a main path, the system multiplies at least a portion of the second frame of image data with the error signal to provide a noise error corrected output signal. In the preferred embodiment, a wavelet filter is used to remove dome shading effects from the frames of image data. In the best mode, the wavelet filter is disposed in the main path and blurred and focused outputs therefrom are weighted, averaged and stored. Coefficients from the weighted, averaged and stored focused frames are compared to coefficients from the weighted, averaged and stored blurred frames to provide a fixed pattern noise error signal. A temporal noise error signal is identified from the weighted, averaged and stored focused frames. The fixed pattern noise error signal and the temporal noise error signals are sparse processed and shunted from a current frame. Thereafter, a constant mean value may be added to provide the output signal. Pixel replacement can be consolidated into a single circuit and positioned prior to the wavelet filter.