A rapid analysis method for cuttings fluorescence images

Abstract

The invention relates to a method for rapidly analyzing a rock debris fluorescence image. The method comprises the following steps that (1) oily component identification training is carried out on the basis of RGB three-dimensional color space, so that a clustering file is obtained; (2) oily component analysis is carried out on the rock debris fluorescence image to be tested according to the clustering file obtained in the step (1). The oily component analysis process specifically comprises the following steps that (2.1) color feature extraction is carried out on the input rock debris fluorescence image to be tested, and a feature vector array is generated; (2.2) each feature vector in the feature vector array is analyzed; (2.3) after analysis work of all the feature vectors in the feature vector array is accomplished, oily components contained in the rock debris fluorescence image and the proportions of the oily components are obtained through calculation. According to a rock debris fluorescence layered clustering method based on non-uniform color space color difference calculation, the oily component identification result and the rock debris fluorescence image analysis result are accurate, oily component identification and rock debris fluorescence image analysis are rapid, the requirement of geological staff of an offshore platform for intelligentized well logging can be met, and therefore the method for rapidly analyzing the rock debris fluorescence image can be widely applied to the geological exploration and development process.

Description

technical field [0001] The present invention relates to a method for analyzing the oiliness of cuttings, in particular to a method for rapid classification of cuttings fluorescence images based on non-uniform color space color difference calculation and color space layered clustering suitable for real-time mud logging on offshore platforms . Background technique [0002] In the process of exploration and development, in order to accurately grasp the geological conditions of different formations, it is necessary to analyze the collected cuttings samples, so as to obtain the lithology and oiliness information of the formations. However, the oiliness contained in cuttings, especially light oil and heavy oil, is usually difficult to observe with the naked eye, so fluorescence logging technology emerged as the times require. This technology is currently widely used in the field of oil and gas layer identification, and is recognized as the best rock chip analysis method. This me...

AI Technical Summary

Problems solved by technology

However, this method has the following problems: 1) The similarity of color features is determined by calculating the color distance, that is, the color difference, but because the RGB color space is not uniform, the clustering results do not match the perception, because the RGB components It not only represents the color, but also represents the degree of lightness and darkness, which leads to the possibility that the distance between similar colors may be large, but the distance between dissimilar colors may be small

In order to solve this problem, someone proposed an improved ISODATA weighted clustering algorithm, which is based on the HSL color space for component identification instead of the RGB color space, avoiding the non-uniformity of the RGB color space, and improving the clustering to a certain extent. However, there is still a small gap with the visual judgment. At the same time, affected by the characteristics of the clustering algorithm, in the case of many classification attributes and a large sample space, the calculation time of ISODATA is too long and it is easy to fall into the local optimum; During sub-training, most methods use the collected fluorescence images as training samples to calculate the number of clusters and cluster centers. When the integrity of the training samples cannot be guaranteed, the component training effect is often unsatisfactory.

Images