Image processing based infrared precursor recognition for NPR cable anchored rock with inclination angles

Recognizing rock fracture precursors is critical for ensuring the safe and stable operation of deep underground engineering. The paper investigates the fracture mechanisms of Negative Poisson’s Ratio material (NPR) and Poisson’s Ratio material (PR) material anchored composite rock specimens (NCR & PCR) with different inclination angles in uniaxial compression test (UCT), monitored with acoustic emission (AE) and infrared (IR) thermography. A new index, Enhanced Infrared Matrix of Damage (EIMD), is defined by processing the noise-suppressed IR temperature matrices with Otsu thresholding, which highlights thermal anomalies associated with crack initiation and growth. On this basis, the correlative index Damage Infrared Energy Response (DIER) is further proposed, serving as a robust indicator of crack-induced radiative energy release. Based on these IR parameters, the integrated isolation forest model is constructed to identify anomaly points during damage evolution, enabling early failure prediction. Experimental results show that NCR exhibits 3.06%–6.2% higher peak strength than PCR across inclination angles of 15°–45°, and NCR-30°/45° retains residual load-bearing capacity through stress curve rebound. AE analyses reveal that PCR undergoes multiple sharp b -value drops in Stage III, with an increasing inclination, whereas NCR maintains a rising b -value and a moderate Low Frequency (LF) ratio, reflecting an enhanced crack resistance ability. EIMD and DIER provide a reliable framework for quantifying rock damage, with anomalies in isolation forest prediction characterized by distinct DIER–EIMD patterns and crack morphologies. The integrated prediction model achieves early warning within 13–51 s (89%–99% peak stress ( σ peak )), with NCR anomalies detected 6–10 s earlier than PCR, demonstrating the stabilizing role of NPR cables in delaying infrared precursors and restraining large-scale crack propagation. This study provides researchers with a new infrared thermography-based perspective for identifying precursors of rock fracture, offering a useful reference for future investigations on predicting the UCT failure behavior of anchored rocks. It also establishes an experimental and theoretical foundation for understanding the mechanical response and failure prediction of anchored composite roof strata in engineering applications.