This study introduces an application of binocular stereo vision technology to achieve high-resolution 3D visualisation of submillimeter discharge channels. The proposed method can capture the depth information of weak luminous discharges and provide 3D discharge trajectories

providing new insights into the discharge mechanism. The camera parameters are selected by a multi-objective optimisation algorithm and calibrated using a micro-chessboard

thus addressing the complex calibration challenges associated with imaging microscopic objects. The brightness threshold-based point cloud filtering algorithm developed for this study improves the imaging quality of luminous channels under low-light conditions. Experimental validation involving 3D reconstruction of complex branched discharge channels in SF6/N2 mixtures reveals the key characteristics of the leader stepping mechanism. The accuracy and feasibility of the proposed method lay a solid foundation for future studies of discharge phenomena and the development of physical models of low-probability breakdown events.