Microfibre reinforcement is widely used for additively manufactured composites, however, it remains a challenging task to resolve the 3D distribution of microfibres in the matrix with low contrast and resolution for image analysis. To address this challenge, we propose a novel iterative destruction–filtering–repairing (DFR) image processing method, and take glass microfibre-embedded extrusion-moulded cementitious filaments (EMCFs) as an example for 3D structural analysis with X-ray computed tomography. Results show that the DFR method enables near pixel-level microfibre diagnosis with high accuracy beyond the ordinary image processing method. The glass microfibres are unevenly distributed in the EMCFs and their orientations are different between the filament body and the interfacial zone, uncovering the mechanisms of extrusion-affected material migration, collision, and redistribution. Our findings make a breakthrough in image analysis with limited resolution and contrast, providing a broad path towards better understanding the microstructure of 3D printed fibre-reinforcement composites.