Measurement for Emulsified Crude Oil Thickness Using Laser-Induced Fluorescence and Visual Imaging Technology

Accurate measurement of emulsified crude oil thickness is crucial for assessing the pollution range and the environmental impact of offshore oil spills. However, the measurement accuracy of conventional Laser-Induced Fluorescence (LIF) is limited by the complex physicochemical properties of emulsified crude oil. To address this issue, a device combining LIF with visual imaging technology was developed to enhance the accuracy of emulsified crude oil thickness measurement. Emulsified crude oil of different thicknesses was observed to exhibit different color features under LIF 405 nm violet laser irradiation. The analysis of the relationship between these color features and oil thickness improved measurement accuracy. In this study, spectral data and images of emulsified crude oil with different thicknesses and oil contents were acquired using the device combining LIF with visual imaging technology. Image features were extracted using a Two-Dimensional Convolutional Neural Network (2DCNN) with an attention mechanism. Partial Least Squares Regression (PLSR) was applied to fit the thickness based on the spectral data and image features. The validation phase, the actual sample for emulsified crude oil with unknown oil content was used. The average relative error of thickness measurement for this sample was 0.759 %. The results showed that the accuracy of the thickness estimation using image features was significantly higher than that obtained spectral data.