Study On Segmentation Of Furnace Flame Based On Fuzzy Clustering

China is a country with less oil and gas than coal, coal consumption accounts for about70%of total energy consumption, the consumption of raw coal accounted for industrial boiler total energy consumption18%in China. However, most of the current system of boiler operation in China stays in the high energy consumption, high emission and low service state, which causes serious pollution to the environment and a huge waste of energy. Because the coal combustion is a complex, strong oxidation reaction, its working condition is unstable and difficult to control. Therefore, how to effectively monitor the burning flame is an important guarantee for the safe, efficient production of coal-fired boiler. The rapid development of image processing and computer vision technology brings the methods and means more intuitive, efficient for boiler combustion state monitoring. Imaging the segmentation technology as one of the core problems of boiler combustion monitoring system based on image processing has been concerned widely. But most of the existing literature on the gray image, the existence of the poor versatility, the segmentation effect and real time is not good enough etc. Aiming at the existing deficiency of boiler flame segmentation algorithm, based on analysis of all kinds of segmentation algorithm based on clustering algorithm has strong versatility, play advantage, combining the advantages of Lab image color space, by improving the traditional fuzzy clustering segmentation algorithm, presents a simulation experiment with fuzzy C means color image segmentation method of hill climbing method and by the method of flame image. The experiment proves that the improved algorithm in general, real time has a significant promotion, and has obtained the very accurate segmentation results.In order to improve the current general flame segmentation algorithm, this paper analyzes the existing imagement segmentation technology, determine the clustering segmentation algorithm as the flame segmentation algorithm. Focus on the study of clustering algorithm, fuzzy clustering algorithm and image segmentation algorithm based on fuzzy clustering of four classical including HCM, FCM and HFCM, and expounds the defects of these exist. Through the study of the use of hill climbing method, the problem of large amount of calculation, the iterative solution is difficult to determine the initial parameters; using Lab color image segmentation effect advantage, solving the problem of poor segmentation results, using after treatment to solve the problem of ignoring the spatial information; boiler flame effectively improve segmentation universality, real-time and effect of segmentation.This paper takes the clustering algorithms used in boiler flame segmentation is one of the main research content The Algorithm effectively improve the flame segmentation universality,adaptability and the segmentation effect, safety, efficient production of the boiler has important application value. Taking the color images as research object, focuses on the research of fuzzy clustering segmentation algorithm, the segmentation of color image after the research has important theoretical significance.