Toward Credible 3D Point-Cloud Learning And Hyperspectral Image Learning: Data Quality, Calibration, and External Validity

Abstract

This review examines a shared methodological problem in 3D point-cloud learning and hyperspectral image learning: how evidence from hierarchical spatial state-space aggregation for point-cloud classification can be placed in analytical dialogue with bidirectional nonlinear spatial-spectral feature learning for hyperspectral images without erasing differences in scale, assumptions, or intended use. Two target papers are triangulated against 12 locally validated publications. The comparison follows neighborhood construction, hierarchy, state-space mixing, sampling robustness, efficiency and deliberately separates mechanistic interpretation from performance ranking, because the latter can conceal incompatible experimental or operational conditions. The synthesis shows that neighborhood construction cannot be interpreted independently of hierarchy, while state-space mixing determines whether an apparent improvement remains meaningful outside the original setting. The strongest claims are therefore those that expose sensitivity, failure conditions, and residual uncertainty. On this basis, the review proposes an auditable pathway from focal mechanism to application claim, with explicit checkpoints for calibration, external validity, and responsible interpretation.