Novel lithology identification method for drilling cuttings under PDC bit condition

Abstract

At present, PDC bit lies in the dominant position in the oil bit market. It is difficult to identify drilling cuttings manually because drilling cuttings produced by the PDC bit are small in quantity and size. To solve the problem, a method for lithologic identification of drilling cuttings based on machine vision is presented. Firstly, the watershed algorithm is used to segment all the drilling cuttings in the original image. After that, oil-bearing drilling cuttings are located by fluorescence images, features of color and texture are extracted from drilling cuttings, a feature library is created and the feature similarity between drilling cuttings and feature library is calculated according to the improved Bhattacharyya distance to complete the lithologic identification of oil-bearing drilling cuttings. Finally, the area proportion of oil-bearing drilling cuttings is calculated in the original image. Experimental results show that the error of the proposed method is the lowest, all of which are less than 5% in accuracy, and the time spent is much lower than that of manual identification (300s), all of which are less than 110s.In image segmentation, the idea of region merging can solve the over-segmentation problem of watershed. In image recognition, a matching method based on feature library is proposed, and the similarity of similar drilling cuttings is increased by improved Bhattacharyya distance, so as to improve the accuracy of the algorithm.