Cross-spectral registration of natural images with SIPCFE
- PDF / 3,764,710 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 55 Downloads / 182 Views
ORIGINAL PAPER
Cross-spectral registration of natural images with SIPCFE Amir Hossein Farzaneh1
· Xiaojun Qi1
Received: 22 January 2018 / Revised: 14 April 2019 / Accepted: 8 January 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Image registration is a viable task in the field of computer vision with many applications. When images are captured under different spectrum conditions, a challenge is imposed on the task of registration. Researchers carefully handcraft a local module insensitive to illumination changes across cross-spectral image pairs to tackle this challenge. We, in this paper, develop an optimized feature-based approach Single Instance Phase Congruency Feature Extractor (SIPCFE) to tackle the problem of natural cross-spectral image registration. SIPCFE uses the phase information of an image pair to quickly identify and describe reliable keypoints that are insensitive to illumination. It then employs a sequence of outlier removal processes to find the matching feature points accurately and the Direct Linear Transformation to estimate the geometric transformation to align the image pair. We extensively study the proposed approach for every module in the system to give more insights into the challenges. We benchmark our proposed method and other state-of-the-art feature-based methods developed for cross-spectral imagery on three datasets with various settings and image contents. The comprehensive analysis of cross-spectral registration results of natural images demonstrates that SIPCFE achieves up to 47.24%, 14.29%, and 12.45% accuracy improvement on the first, second, and third dataset, respectively, over the second best registration method in the benchmark. Keywords Cross-spectral registration · Phase congruency · Near infrared · Feature-based image registration
1 Introduction A cross-spectral image pair is a pair of two corresponding images captured in different imaging configurations such as different camera exposures, different camera positions, and different sensors. These different configurations make the images in one pair not perfectly aligned; hence, registering them is a challenging task in computer vision applications. When registering two images, the aim is to find a geometric transformation between a pair of corresponding images to compensate for the rotation, translation, and scaling differences. The transformation is then used to spatially align, superimpose, or match the images in a pair. With two registered images, it is easier to fuse information or describe the differences between them. Cross-spectral image registration has broad applications in remote sensing, object detection, noise reduction, 3D image reconstruction,
B
Amir Hossein Farzaneh [email protected] Xiaojun Qi [email protected]
1
Department of Computer Science, Utah State University, 4205 Old Main Hill, Logan, UT 84322-4205, USA
image fusion, video surveillance, medical image analysis, and image mosaicking. In this paper, we focus on registering the RGB spectrum and near-infr
Data Loading...
