Research On The Flame Image Edge Detection And Restoration Algorithm

There is direct relationship between boiler combustion stability, sufficiency and coal-fired power plant operation stability, economy, which urged the occurrence of various boiler flame detectors. But all these detectors must be fixed to directly touch with flame inside furnace, which is required high-quality material but low-accurate test results. Therefore, there is emerged flame image detection technology.Generally, the flame image detection process is included:extracting flame image information, drawing flame feature, identifying detection and judging results. The furnace inside area is divided into pulverized coal no-combustion zone, preliminary combustion zone, fully combustion zone, flame trail and flue gas area, which all can be shown in flame image detection results. Based on flame image detecting test results, precisely identify different combustion area and analyze combustion flame characteristics, which should be focus on image edge detail and put forward flame image edge detection and restoration. Through the image edge detection and restoration technology, improving flame image information, clearly seeing the different combustion area distribution, providing powerful evidence for determining pulverized coal burning, ensure the boiler operation safe, stable, efficient and economic.This paper, firstly instructing the research background and importance of flame image detection and restoration work, detailed describing the image edge detection technology and restoration technology development, and then showing the overall paper content structure. Based on comparing different classical image edge detection algorithms, propose gray leading filter algorithm from the test image filtering algorithm perspective. Combining the gray leading filtering algorithm and better Prewitt operator, present the gray leading filtering image edge detection algorithm, which is innovatively raised the algorithm filter window top center reference point, correction factor, and calculated weight filter space according to the different window pixel gray values, stayed the flame image edge information when removing the image noise. After filtering, image edge information is got by improving Prewitt gradient operator, in which extending the operating template size to3x3, expanding the gradient direction to eight, to overcome the defect of furnace flame image edge irregular.There always have to miss some edge detection detail for flame image by existing edge detection algorithm due to the flame is surrounded by smoke in the real situation. According to the problem of image edge detection results, this paper proposes an improved image edge repair method of triangular area. By using this image edge repair method, we can optimization the detective image obtained by the image edge detection algorithms. Combine the proposed image edge detection algorithm based on gray dominant filtering with improved triangular area image edge repair method together, which is indicated that the image edge detection result tends to be ideal.