New unsupervised method segments low-contrast concrete XCT images

Researchers have developed a novel unsupervised approach for segmenting low-contrast X-ray Computed Tomography (XCT) images of concrete. This method addresses the challenge of similar X-ray attenuation coefficients between aggregates and mortar, which typically prevents intensity-dependent segmentation. The technique utilizes a self-annotation process that combines superpixel algorithms with the receptive field of a Convolutional Neural Network (CNN) to learn global-local relationships within the images. Evaluated against manually annotated data, this methodology demonstrated superior performance compared to simple greyscale thresholding, achieving a better balance of sensitivity and precision for identifying concrete aggregates. AI

IMPACT This unsupervised segmentation technique could reduce the need for manual annotation in material science research, accelerating analysis of concrete structures.

RANK_REASON The cluster contains an academic paper detailing a new methodology for image segmentation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• CNN
• X-Ray Computed Tomography
• concrete
• vision transformers

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →