Abstract: Herein, a rock mass information system for tunnel boring machine (TBM) tunneling faces based on machine vision technology is developed to enable safe and efficient image acquisition and intelligent identification of rock mass information at the tunneling face. The system comprises three modules: image acquisition, control and communication, and image processing and visualization. An anti-vibration, sealed housing is developed to protect the image acquisition module, enabling stable installation inside the cutterhead manhole. In addition, a multi-linear-array camera collaborative rotation acquisition method is proposed to achieve wide-area image capture within the confined space in front of the cutterhead. A hybrid wired-wireless transmission scheme is employed to ensure stable system communication. To address image distortion and misalignment, a multi-angle image processing method based on error correction is proposed. Corresponding software is also developed to enable acquisition control and data visualization. Validation through laboratory and on-site tests confirms the following results: (1) the system provides an image pixel equivalent better than 0.1 cm/pixel at a shooting distance not exceeding 950 mm; (2) the designed terminal structure satisfies practical engineering requirements for vibration resistance and sealing, demonstrating stable operation under TBM construction conditions and reliable acquisition of tunnel face images; and (3) during field deployment, the system completes a single acquisition cycle in approximately 280 s, with an average data transmission rate of about 80 Mbps and a packet loss rate below 1%, thereby ensuring the acquisition of clear two-dimensional tunnel face images.

Key words: TBM, tunnel face, machine vision, two-dimensional image, rock mass information