The study of turbulence characteristics in two-phase flows always remained challenging due to its chaotic nature. The hidden coherent structures are dynamically important and play significant role to define their macro-characteristics. Here, we present Linear Stochastic Estimation (LSE)-based approach combined with Proper Orthogonal Decomposition (POD) to perform analysis of two-dimensional slug flow. Analysis is performed in the film region of slug flow which is analogous to stratified flow. The instantaneous velocity of both phases is obtained experimentally by Particle Image Velocimetry (PIV) and Laser-Induced Fluorescence (LIF). The goal is to identify the dominant flow patterns from both phases and reconstruct the flow field. Reference signals from POD eigenmodes and fluctuating velocity are used for LSE estimation of velocity field. The impact of nature of reference signal, location of reference lines and number of reference points on the estimation is also evaluated based on the correlation values. It is observed that two reference lines based LSE estimated velocity identified all dominant flow patterns from both phases of slug flow. Examination of spatial contents of the estimated velocity shows good qualitative agreement with the POD modes at all scales. It is also noted that the location of the reference lines identified the flow motions not only near the specified reference lines but also in the overall field. This combination of LSE and POD is therefore an improvement over the existing LSE approach and can be used in variety of flows for reconstruction even at small scales and could possibly lead to notable advancements in accurate prognosis of multiphase flow regimes. © 2023, The Author(s), under exclusive licence to The Brazilian Society of Mechanical Sciences and Engineering.