Real-time adaptive visible and infrared image registration based on morphological gradient and C_SIFT
- PDF / 2,406,440 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 12 Downloads / 175 Views
ORIGINAL RESEARCH PAPER
Real-time adaptive visible and infrared image registration based on morphological gradient and C_SIFT Qiang Zeng1 · Jianhua Adu1 · Jiexin Liu1 · Jianxing Yang1 · Yuanping Xu1 · Mei Gong1 Received: 27 July 2018 / Accepted: 16 February 2019 © Springer-Verlag GmbH Germany, part of Springer Nature 2019
Abstract Since the visible and infrared images have different imaging mechanisms, the difficulty of image registration has greatly increased. The grayscale difference between visible and infrared images is very disadvantageous for extracting feature points in homogenous region, but they both retain the obvious contour edge in the scene. After using the morphological gradient method, the grayscale edge of visible and infrared images can be obtained and their similarity is greatly improved, and their difference may be seen as the difference in brightness or grayscale. Therefore, we proposed a novel algorithm to realise real-time adaptive registration of visible and infrared images using morphological gradient and C_SIFT. Firstly, the morphological gradient method is used to extract the rough edges of visible and infrared images for aligning their visual features as a single similar type. Secondly, the C_SIFT feature detection operator is used to detect and extract feature points from the extracted edges. The C_SIFT uses the centroid method to describe the main direction of feature points, makes rotation invariance feasible. Finally, to verify the effectiveness of the proposed algorithm, we carried out a series of experiments in eight various scenarios. The experimental results show that the proposed algorithm has achieved good experimental results. The registration of visible and infrared images can be completed quickly by the proposed algorithm, and the registration accuracy is satisfactory. Keywords Visible image · Infrared image · Registration · Morphological gradient · SIFT · Real-time
1 Introduction The use of multi-modal sensors to acquire images for registration and fusionis is of special significance for the development of computer vision. Due to the different imaging mechanisms of multi-modal sensors, the acquired image information is complementary. For example, infrared image has outstanding detection ability at night and good antiblocking ability, and visible image has high resolution and retains rich scene information. Compared with single-mode images, multi-modal images can reflect the different imaging characteristics of the scene or objects from different aspects through different sensors, so that the obtained information is more reliable and complementary. Infrared imaging sensors have excellent night detection capabilities, fog penetration, and anti-blocking capabilities, which are typically * Jianhua Adu [email protected] 1
Software Department, Chengdu University of Information Technology, Chengdu 610225, China
complementary to the high-resolution detailed information retained by visible imaging sensors. Therefore, visible and infrared image become more common combination in m
Data Loading...
