WAVELETS FOR NON-METRIC CAMERA SELF-CALIBRATION

Jun-Fu Ye*, Jaan-Rong Tsay, Dieter Fritsch

Abstract

This study evaluates the developed Wavelet Additional Parameter (WAP) models for non-metric camera self-calibration under diverse control and navigation constraints. Empirical results from the 24 test cases demonstrate that the developed WAPs achieve RMSD(XY) and RMSD(Z) reductions of 78.8%–86.3% and 83.6%–95.6%, respectively, relative to the uncompensated cases. Furthermore, these WAPs successfully mitigate lens distortions without underfitting through their spatial localization and orthogonality. Under the Type 2 configuration (sparse GCPs supplemented with navigation data), the developed models exhibit superior internal consistency, outperforming the traditional BINGO polynomial model by 33.3%–38.4% and 30.7%–36.0% in terms of tie-point a posteriori precisions in the RMS(σₓᵧ) and RMS(σ????), respectively. By leveraging mathematically orthogonal wavelet bases and an adaptive screening workflow to prevent overparameterization, this approach ensures high numerical stability and effectively bridges the gap between low-cost sensors and high-precision UAV-based aerial surveying.

Keywords: Bundle adjustment, Wavelet Additional Parameters (WAPs), Camera self-calibration, Systematic errors

Full Text: