Abstract: The variability in the scale and spatial configuration of structural planes introduces significant uncertainty in preventing key block collapses and optimizing targeted support design for tunnels. In this study, we developed a three-dimensional extension algorithm for structural planes based on two-dimensional trace features extracted from sequential tunnel face images. By integrating this algorithm with multi-scale modeling, we established an equivalent rock mass model retaining only the large-scale controlling structural planes. We analyzed the spatial interactions between key blocks and tunnel faces and, using a block-search algorithm, identified the geometric characteristics of surrounding key blocks. Probabilistic models describing block height and volume distributions were then constructed, and their spatial patterns—size, shape, and orientation—were clarified through rose diagrams. Taking the Fanshishan Tunnel on the Dongliang Expressway in Shandong Province as a case study, a targeted rock-bolt support scheme considering key block stability was proposed and compared with conventional support. Results show that under targeted support, the maximum surrounding-rock displacement decreased from 99.17 mm to 26.96 mm, while the peak bolt axial force dropped from 1.3×10³ kN to 6.5×10² kN, a reduction of about 50%. These findings demonstrate that targeted support effectively mitigates local deformation and stress concentration induced by key block instability, thereby enhancing tunnel stability and optimizing support resource allocation. The proposed method provides a theoretical and practical reference for stability analysis and support design in structurally controlled jointed rock mass tunnels.