Edge-Enhanced Super-Resolution Reconstruction of Rock CT Images

Chennian Gao, Chen Qiu

Abstract: Computed Tomography (CT) technology for rock geological research has been widely applied due to its advantages of three-dimensional imaging, non-invasiveness, and easy integration with computer simulations. However, due to economic costs and time constraints, acquiring high-resolution CT images of rocks with a larger field of view has become increasingly challenging. Therefore, characterizing the pore structure features of rocks is a challenging and difficult task. To address this issue, this paper proposes a super-resolution reconstruction of rock CT images based on spatially-adaptive feature modulation for efficient image super-resolution (SAFMN), trainable Sobel convolution, and bicubic enhancement. The reconstruction effectiveness of the two-dimensional images is evaluated through quantitative extraction and qualitative visualization. Experimental results demonstrate that the improved model performs well under different upscaling factors, outperforming the original SAFMN model and some existing convolutional neural network-based super-resolution methods. Our source codes are available at: https://github.com/gao131/Edge_SAFMN.