Biologically-Inspired Empirical Mode Decomposition Fusion Of Visible And Infrared Images
No Thumbnail Available
Links to Files
Permanent Link
Collections
Author/Creator
Author/Creator ORCID
Date
2012
Type of Work
Department
Electrical and Computer Engineering
Program
Doctor of Engineering
Citation of Original Publication
Rights
This item is made available by Morgan State University for personal, educational, and research purposes in accordance with Title 17 of the U.S. Copyright Law. Other uses may require permission from the copyright owner.
Abstract
Visible and infrared image fusion technology effectively enhances image information content and rendering quality. They are many image fusion approaches that can be utilized to produce high quality fused images. Outdoor scenes exhibit no linearity, are random, and their images have independently distributed pixel colors. Many of the existing image fusion methods including but not limited to, Principal Component Analysis, Wavelet Transform based, make assumption of both image content linearity and dependent distribution and when applied to grayscale fusion. Empirical Mode Decomposition (EMD) represents input images as Instrinsic Mode Functions (IMFs) carrying their spatial and frequency components about each pixel with no a priori assumption. Visible and infrared signals lay in different spectrum ranges. Spectral emissivity and reflectivity, depending on wavelength, special contrast can be enhanced not only in visible video but also in infrared imagery. This work proposed a biologically-inspired fusion of visible and infrared images based on EMD and opponent processing (Bio-EMD). First, registered visible and infrared captures of the same scene are decomposed into Intrinsic Mode Functions through EMD. Then, the fused image is generated by opponent processing the source IMFs. Finally, the results are evaluated based on the amount on information transferred, the clarity of details, vividness of depictions, and range of meaning differences in lightness and chromaticity. Perceptual comparison of the results showed that opponent-processing-based techniques outperformed algorithms based on intensities, and some other techniques. Quantitative assessment confirmed that the proposed technique transferred twice information as much as the state-of-the-art methods did. Bio-EMD provided a high level sharpness, yielded more natural-looking colors, and produced a high magnitude of visually meaningful differences in lightness and chromaticity for the fusion of low-light visible and infrared images. These results were obtained without optimization of filters involved in the process.