Two-phase flow characterization in plate heat exchangers has traditionally relied on visual classification methods that lack objectivity and reproducibility, particularly for complex slug-like flow patterns. In this study, a novel quantitative framework using filter-based segmentation techniques to automatically analyse gas-liquid flow behaviour in corrugated channels with different chevron angles (β=30^∘, 45^∘, 60^∘) under realistic process conditions is developed. High-speed imaging combined with dual filtering approaches enabled reliable detection of gas structures and their classification into six clusters based on geometric and morphometric features (e.g., equivalent diameter, circularity, aspect ratio, and complexity score). Although the study encompasses Reynolds numbers up to 5000, quantitative investigations were restricted to the range 850 ≤ Re ≤ 2200 where the detection techniques are sufficiently robust for quantitative analysis. Experimental investigation reveals that chevron angle primarily controls spatial gas phase distribution: β=60^∘ configurations concentrate gas flow centrally (up to 77.5%) promoting wavy longitudinal flow, while β=30^∘ maintains balanced cross-flow distribution across channel width. Higher Reynolds numbers consistently drive transitions from slug-like flow to dispersed bubbly flow, with different cluster velocities beginning to homogenize at elevated Reynolds numbers. Gas fraction enhances coalescence processes, with irregular gas structures dominating 70.0−79.0% of the total gas area likely due to turbulent conditions. Volume-based gas fraction estimation systematically exceeds area-based methods by accounting for actual plate geometry where large structures occupy deep channel regions. The framework enables objective quantification of complex flow phenomena, providing enhanced understanding of bubble dynamics and morphological evolution in corrugated channels beyond traditional visual classification methods.