Repository hosted by TU Delft Library

Home · Contact · About · Disclaimer ·

Small maritime target detection through false color fusion

Author: Toet, A. · Wu, T.
Institution: TNO Defensie en Veiligheid
Source:DeWeert, M.J.Saito, T.T., Optics and Photonics in Global Homeland Security IV, 17 March 2008 through 20 March 2008, Orlando, FL, 6945
Proceedings of SPIE - The International Society for Optical Engineering
Identifier: 22171
doi: doi:/10.1117/12.773279
ISBN: 9780819471369
Article number: 69450V
Keywords: Vision · Colour · image fusion · target detection · Clutter · False color fusion · Image fusion · Small targets · Target detection · Algorithms · Boolean functions · Computer networks · Evolutionary algorithms · Frequency bands · Image segmentation · Imaging systems · Imaging techniques · Mathematical transformations · Optical properties · Optics · Optoelectronic devices · Photonics · Regression analysis · Scheduling algorithms · Security systems · Targets · color imaging · False color · Far-infrared frequency · First order statistics · Floating mines · Homeland security · IR bands · Maritime operations · Morphological opening · Multi bands · Multi-spectral · Search and rescue operations · Small targets · Target detections · target signatures · Uncorrelated noise · visual imaging · Iridium


We present an algorithm that produces a fused false color representation of a combined multiband IR and visual imaging system for maritime applications. Multispectral IR imaging techniques are increasingly deployed in maritime operations, to detect floating mines or to find small dinghies and swimmers during search and rescue operations. However, maritime backgrounds usually contain a large amount of clutter that severely hampers the detection of small targets. Our new algorithm deploys the correlation between the target signatures in two different IR frequency bands (3-5 and 8-12 ìm) to construct a fused IR image with a reduced amount of clutter. The fused IR image is then combined with a visual image in a false color RGB representation for display to a human operator. The algorithm works as follows. First, both individual IR bands are filtered with a morphological opening top-hat transform to extract small details. Second, a common image is extracted from the two filtered IR bands, and assigned to the red channel of an RGB image. Regions of interest that appear in both IR bands remain in this common image, while most uncorrelated noise details are filtered out. Third, the visual band is assigned to the green channel and, after multiplication with a constant (typically 1.6) also to the blue channel. Fourth, the brightness and colors of this intermediate false color image are renormalized by adjusting its first order statistics to those of a representative reference scene. The result of these four steps is a fused color image, with naturalistic colors (bluish sky and grayish water), in which small targets are clearly visible. Keywords: target detection; false color fusion, small targets, clutter, image fusion