How can neural networks improve the uniformity of imaging array detectors? This paper introduces two neural network-based methods for nonuniformity correction in imaging array detectors. The methods exploit image properties to compensate for calibration deficiencies and maximize output entropy. The first method uses a self-organizing net to produce a linear correction with continuously adapting coefficients. The second employs a contrast equalization curve to match pixel distributions. Although originating from silicon detectors, the approach is broadly applicable to various array detectors used in infrared imaging and high-energy physics. It will be most useful in the area of **computer science**.
This paper in Neural Computation investigates the application of neural networks to improve the performance of imaging systems. The interdisciplinary nature of this work aligns with the journal's focus on computational approaches to neuroscience.