Super resolution reconstruction of μ-CT image of rock sample using neighbour embedding algorithm

• Published in: Physica A Vol. 493; pp. 177 - 188
• Main Authors: Wang, Yuzhu, Rahman, Sheik S., Arns, Christoph H.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2018
• Subjects: [formula omitted]-CT; Neighbour embedding; Self-similarity; Super resolution; μ-CT; Neighbour embedding; Super resolution; Self-similarity
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2017.10.022

X-ray computed tomography (μ-CT) is considered to be the most effective way to obtain the inner structure of rock sample without destructions. However, its limited resolution hampers its ability to probe sub-micro structures which is critical for flow transportation of rock sample. In this study, we propose an innovative methodology to improve the resolution of μ-CT image using neighbour embedding algorithm where low frequency information is provided by μ-CT image itself while high frequency information is supplemented by high resolution scanning electron microscopy (SEM) image. In order to obtain prior for reconstruction, a large number of image patch pairs contain high- and low- image patches are extracted from the Gaussian image pyramid generated by SEM image. These image patch pairs contain abundant information about tomographic evolution of local porous structures under different resolution spaces. Relying on the assumption of self-similarity of porous structure, this prior information can be used to supervise the reconstruction of high resolution μ-CT image effectively. The experimental results show that the proposed method is able to achieve the state-of-the-art performance. •An image-embedding method is applied to improve the resolution of μ-CT.•High resolution SEM image is used to extract high- and low- image patch pairs.•Image patch pairs are used in supervise the reconstruction of μ-CT image.