Precise measurement of warp and weft thread density is critical for fabric quality control and process monitoring. Conventional spectral and orientation-based methods suffer from scale ambiguity, noise sensitivity, and inconsistent performance across fabric types. This study proposes a physically interpretable multi-domain integration framework for warp–weft count estimation by fusing FFT-derived spectral features, Gabor-based directional texture energy, and wavelet-based multi-scale decomposition. Four integration strategies are introduced: weighted physical integration, confidence-gated integration, ratio-consistency integra tion, and energy-normalised integration. The ratio-consistency method enforces a novel structural constraint whereby the final weft count is derived as ˆH = ˆ W/Rfinal, enforcing proportional coherence between orthogonal yarn directions by design. The proposed methods are fully data-driven, require no sensor-specific calibration, and explicitly model measurement instability. Experimental evaluation on a corpus of 616 image regions derived from 7 woven fabric samples (plain, twill, and satin; 10 80threadscm−1; 88 non-overlapping 1200×1200px crops per sample on average) demonstrates that ratio-consistency integration achieves the lowest interquartile range (warp IQR=5.88, weft IQR=6.42) and the highest entropy ratio (3.050, defined as Hafter/Hbefore over per-domain energy weights) among all evaluated strategies. Inter-method consistency is validated against Gabor derived reference counts. The framework is suitable for industrial vision systems requiring deterministic, explainable, and stable warp–weft density measurement.