Dynamic range reduction and contrast adjustment of infrared images in surveillance scenarios

The high thermal sensitivity of modern infrared (IR) cameras allows us to distinguish objects with small temperature variations. In comparison with the dynamics of standard displays, the sensed IR images have a high dynamic range (HDR). In this context, suitable techniques to display HDR images are required in order to improve the visibility of the details without introducing distortions. In the recent literature of IR image processing, a common framework to perform HDR image visualization relies on DR reduction (DRR) with a cascaded processing for local contrast adjustment (CA). In this work, a novel method, named cluster-based DRR and contrast adjustment (CDCA) is introduced for the visualization of IR images. The CDCA method is composed of two cascaded steps: (1) DRR clustering-based approach and (2) a CA module specifically designed to account for IR image features. The effectiveness of the introduced technique is analyzed using IR images of surveillance scenarios collected in different operating conditions. The results are compared with those given by other IR-HDR visualization methods and show the benefits of the proposed CDCA in terms of details enhancement, robustness against the horizon effect and presence of hot objects. (C) 2013 Society of Photo-Optical Instrumentation Engineers.