Carbonate microfacies study by using images processing algorithms, K-mean clustering and nearest neighbor segmented classifying: an example from the Salman Oil and Gas Field, Persian Gulf, Iran

Finding and quantifying microscopic features such as matrix and grains, fabrics, porosity, fossil contents and diagenesis are crucial to improving the results of a microfacies study. Moreover, the application of image processing seems essential in analysis of hydrocarbon fields. There is a wide range of available image processing algorithms. However, these algorithms are dealing with many difficulties when faced with complex microfacies study objectives.In this paper, 170 thin section photographs from a Permo-Terias formation of Salman field in South-west of Iran were analyzed. Using the suggested histogram equalization algorithm, the selected thin section images were improved in a way to be comparable with the reference photographs. Afterward, the main microfacies major features such as matrix texture, boundaries, fossil content and appearance are characterized by applying functional image processing algorithms and sensitivity analysis of the algorithm results. Accurate grain size is measured in a designed Graphical User Interface (GUI). Next, pore detection and 2D porosity values are calculated by K-means clustering of A and B parameters in L*A*B color image space. Finally, different minerals in the matrix, cement, and porosity are classified and distribution of them are visualized and plotted on a scatter plot to determine the exact facies types.