Pose-invariant calibration via SVD-based homography for extended line-structured light scanning

State-of-the-art calibration methods for line-structured light systems typically employ two main approaches. Sequential parameter estimation often suffers from cumulative error propagation, as inaccuracies in the initial camera calibration inevitably cascade into the light plane estimation. Moreover, this approach is highly sensitive to image noise. Alternatively, while direct whole-system fitting circumvents this drift, it relies on predefined geometric configurations and rigid mechanical constraints that are frequently unattainable or cost-prohibitive. Specifically, this method requires perfect mechanical orthogonality between the line-structured light plane and the travel vector; achieving this is mechanically costly, and misalignments introduce systematic cross-axis shear. To fill this gap, this paper introduces a practical integration of homography mapping, Singular Value Decomposition (SVD)-based orthogonal consensus, and encoder-coupled longitudinal reconstruction. The proposed engineering advancement recovers a global consensus plane that compensates for mechanical tilt, allowing arbitrary system orientations without precise alignment. By combining continuous encoder feedback with an iterative pulse-to-millimeter refinement loop, the system supports extended longitudinal reconstruction; while practically constrained by the physical tolerances of the motion hardware, this encoder-driven architecture bypasses the spatial limits of fixed-stage calibration. This effectively minimizes the “stitching” artifacts and cumulative errors common in traditional fixed-stage methods. Experimental validation against both a precision-stepped target and certified gauge blocks shows that measurement accuracy remains consistent throughout the scanning volume, confirming the significant reduction of systematic cross-axis coupling under mechanical misalignment.