Fast Adaptive Nonuniformity Correction for Infrared Focal-Plane Array Detectors
- PDF / 1,880,643 Bytes
- 11 Pages / 600 x 792 pts Page_size
- 37 Downloads / 240 Views
Fast Adaptive Nonuniformity Correction for Infrared Focal-Plane Array Detectors Esteban Vera Department of Electrical Engineering, University of Concepcion, Casilla 160-C, Correo 3, Concepcion, Chile Email: [email protected]
Sergio Torres Department of Electrical Engineering, University of Concepcion, Casilla 160-C, Correo 3, Concepcion, Chile Email: [email protected] Received 29 October 2003; Revised 15 January 2005 A novel adaptive scene-based nonuniformity correction technique is presented. The technique simultaneously estimates detector parameters and performs the nonuniformity correction based on the retina-like neural network approach. The proposed method includes the use of an adaptive learning rate rule in the gain and offset parameter estimation process. This learning rate rule, together with a reduction in the averaging window size used for the parameter estimation, may provide an efficient implementation that should increase the original method’s scene-based ability to estimate the fixed-pattern noise. The performance of the proposed algorithm is then evaluated with infrared image sequences with simulated and real fixed-pattern noise. The results show a significative faster and more reliable fixed-pattern noise reduction, tracking the parameters drift, and presenting a good adaptability to scene changes and nonuniformity conditions. Keywords and phrases: infrared detectors, focal-plane array, nonuniformity correction, fixed-pattern noise, neural networks, least mean square.
1.
INTRODUCTION
Infrared detectors are widely used in a variety of applications such as defense, surveillance, remote sensing, and astronomy. Usually, the infrared imaging sensors are based on the infrared focal-plane array (IRFPA) technology [1, 2, 3]. An IRFPA can be considered as an array of independent detectors aligned at the focal plane of the imaging system. Unfortunately, every detector on the IRFPA can have different responses under the same stimulus, what is considered as the nonuniformity problem, leading then to the presence of a fixed-pattern noise (FPN) in the resulting images. Furthermore, this FPN presents a noticeable temporal drift, which is even severe in uncooled infrared systems. Under this scope, nonuniformity correction (NUC) is a necessary and unavoidable task to be performed in order to achieve higher-quality infrared images, or image sequences, eliminating thus the unwanted FPN. In this way, NUC techniques normally assume a linear model for the detectors, characterizing the nonunifomity response problem as a gain and offset estimation problem per detector. The most accurate NUC methods are based on the use of uniform infrared sources, where the simplest and most used one is the two-point calibration [4], which employs two
blackbodies at different temperatures to calculate the exact gain and offset of each detector on the IRFPA through the use of a simple line fitting procedure. However, these referencebased NUC methods must be often repeated in order to follow the temporal drift of the nonuniformity characteris
Data Loading...
