Research And Application Of Automatic Image Recognition Technology For Maceral Groups In Bituminous Coal

The physical and chemical properties as well as coking properties of coal are determined by coalification degree and macerals.The automatic determination of vitrinite reflectance and macerals of coal can not only solve the problems of heavy labor,low efficiency and the deviation from traditional manual determination,but also push the application of parameter index of coal petrology to be used in coal classification and coking coal blending industrial production.Numerous research works have been done home and abroad in automatic determination of coal petrology over the past decades.However,due to the complexity of maceral composition and its change,there are still many technological difficulties unsolved in the automatic recognition of macerals.This research combines theories,methods and technologies about coal geology,coal petrology,mechanic manufacturing and automatic control,digital image processing and computer software development.The research is focused on the interdisciplinary work involving development of automatic scanning and focusing device for reflective microscope;optimization of influencing factors and key collection index for the qualities of micro-image of coal petrology;preprocessing of micro-image of coal petrology.Research is also done on the methods and technology on micro-image automatic collection system of coal petrology and the automatic recognition of maceral group and related software development.Testing and verification have been done to representative coal samples from different ranks in China,using the research results developed in this work.The following results have been obtained:(1)The microscope-based hardware platform for micro-image automatic collection is developed,which includes the development of auto-scanning and auto-focusing device based on reflective microscope;and technology of automatic focus of micro-image of coal petrology based on multi-window focus evaluation and search.It enables the automatic fast and efficient collection and storage of micro-image of coal petrology.(2)Major control factors are discovered,which influence the collection quality of micro-image of coal petrology.And optimization method is raised in getting the preferred index about light intensity and exposure time for micro-image collection of coal petrology for coals with different degrees of coalification.Mathematical model is built thereafter.(3)An effective method for removing micro-images of resin zone is proposed based on the mixed feature multi-characteristics extraction method.Gray abnormal boundary between different macerals in micro-image of coal petrology is discovered.The cause of its formation is disclosed.Based on Prewitt operator,a removal technique of gray abnormal boundary of gray characteristics of maceral is formed.Technology of preprocessing micro-image of coal petrology is developed.(4)For the first time,automatic image segmentation and recognition algorithm for GKMC maceral of coal petrology is developed,namely A Gaussian K-means clustering algorithm for image automatic segmentation and recognition technology of maceral groups is developed.Programing is done by the use of Matlab software.(5)Automatic image testing technology is innovated and the device system for GKMC maceral of coal petrology is created with independent intellectual property rights.Using the device system developed in this work,micro-images of 45 representative coal samples from different rank in China are automatically collected and analyzed.The results are compared with the standard results of manual identification done by domestic authoritative experts in coal petrology field.The results show that the average range are 2.3%,2.3%and 1.5%,respectively for vitrinite,inertinite and liptinite group determined by the two methods.According to the accuracy analysis method in national standard GB/T 18510-2001,the statistics of three maceral groups tc<tt,the critical value of free degree under the condition of(n-1),showing that the analysis results have no bias with manual identification data.It can prove that the device system developed in this research can realize the automatic testing of maceral of coal petrology with speed and efficiency.The device system has good repeatability and high accuracy.