A method for characterizing liquid samples with a thin mesh using infrared spectroscopic imaging is presented. Through decomposition and residual interpolation, the baseline distribution is determined. The spectra are then optimized iteratively to obtain corrected spectra and residuals for each component. This method overcomes water signal interference and enables differentiation of interaction types. It is applicable to extract compound interaction information in water environments and beyond.