Extraction Of Gas Ash Object Based On Image

As a by-product of blast furnace smelting,gas ash contains large amounts of carbonaceous matter and metallic elements,such as iron,lead,zinc,as well as other alkali metal oxides.The content of unburned coal and coke can reflect the rationality of the ingredient of blast furnace.The circulated enrichment of zinc,aluminium,copper and other impurities can also impact on the recycle of gas ash.Therefore,quantitative analysis of the composition of the gas ash is of great significance for guidance of blast furnace production and the optimization of the use of gas.The extraction of unburned coal,coke and metal oxide in the gas ash is the prerequisite for the automatic analysis of the ash composition.According to the differences of color and texture of different components in the gas ash,microscopic images are segmented with K-means and ISODATA clustering algorithm,and a Mean Shift clustering method combing spatial and chromaticity field is proposed to cluster these object areas,and achieved acceptability results.The main contributions of this dissertation is as follows:(1).Based on referring to relevant literature,the research status of the comprehensive utilization of gas ash,the image segmentation algorithm,and the background and significance of the research are summarized.(2)The features of color and texture of the gas ash microscopic image are analyzed,six color component information based on RGB and LUV color models,5 texture related features as energy,entropy,moment,local smooth,maximum probability based on gray level cooccurrence matrix and 6 intensity related features as contrast,mean,standard deviation,skewness,uniformity,peak and kurtosis based on gray-level statistics are extracted.The differences and distinguishing of features between the different components of the gas ash were analyzed,which provided the basis for subsequent object extraction and identification.(3)Object extraction scheme is designed.With algorithms of K-means and the ISODATA,the object areas in the typical gas ash microscopic images are clustered,and a Mean Shift clustering method combing spatial and chromaticity field is proposed to cluster these object areas,acceptability result is obtained.(4)An image processing framework is built with Visual C++ 6.0 and OpenCV,and algorithms of feature extraction and object extraction employed in this dissertation are programmed and implemented,which offers a software platform for auto-analysis and recognition of different components in gas ash.The special and innovations of the dissertation is as follows: the components gas ash are analyzed with digital image processing technology;The gas ash microscopic images is effectively segmented by an mean shift clustering algorithm combining space and color gamut;A Mean Shift clustering method which combines spatial and chromaticity field is proposed to cluster the object areas of the gas ash microscopic image;Combining the visual features of color and texture,the object areas of the carbonaceous matter in the gas ash microscopic image are extracted.