Infrared polarization and intensity image fusion based on bivariate BEMD and sparse representation
- PDF / 1,764,862 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 31 Downloads / 222 Views
Infrared polarization and intensity image fusion based on bivariate BEMD and sparse representation Pan Zhu 1,2
& Lu Liu
1,2
& Xinglin Zhou
1,2
Received: 6 November 2019 / Revised: 27 August 2020 / Accepted: 9 September 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The issue of infrared polarization and intensity images fusion has shown important value in both military and civilian areas. In this paper, a novel fusion approach is addressed by reasonably integrating the common and innovation features between the above two patterns of images, employing Bivariate Bidimensional Empirical Mode Decomposition (B-BEMD) and Sparse Representation (SR) together. Firstly, the high and low frequency components of source images are separated by B-BEMD, and the “max-absolute” rule is used as the activity level measurement to merge the high frequency components in order to effectively retain the details of the source images. Then, the common and innovation features between low frequency components are extracted by the tactfully designed SRbased method, and are combined respectively by the proper fusion rules for the sake of highlighting the common features and reserving the innovation features. Finally, the inverse B-BEMD is performed to reconstruct the fused image. Experimental results indicate the effectiveness of the proposed algorithm compared with traditional MSTand SR-based methods in both aspects of subjective visual and objective performance. Keywords Image fusion . Sparse representation . B-BEMD . Common and innovation features
1 Introduction Owing to the distinct imaging mechanism, infrared polarization and intensity images can reflect different features, and each other includes complementary and discriminative information from the same scene. For example, infrared intensity image can reveal the obvious target
* Pan Zhu [email protected]
1
Key Laboratory of Metallurgical Equipment and Control Technology, Ministry of Education|, Wuhan University of Science and Technology, Wuhan 430081, China
2
Hubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan 430081, China
Multimedia Tools and Applications
from background information, whereas, the target often has low contrast, fuzzy details and unclear contour when the temperature difference between target and background is unconspicuous [14, 16]. Interestingly, infrared polarization image can well reflect the boundary, target and contour information of the target depending on their polarization information versus the background [29–31]. However, infrared polarization image cannot present abundant background information [29].The polarization image is obtained by calculating the corresponding specific intensity image, which is anticipated that this pair of source images possess common and innovation structure features [22, 37]. In the result, the traditional fusion algorithm will lead to the images integration quality decline on account of innovation featu
Data Loading...
