The Research Of Recognition Method Of The New Ignition Based On The Image Technology

The damaged heating pipe of tubular furnace cause leakage of tar combustion,which not only cause pipeline further damaged, caused incalculable economic losses,but also a large amount of harmful gases from the burning of tar, seriously pollute theecological environment. Therefore, in order to reduce the huge economic loss andenvironmental pollution, the recognition of abnormal combustion flame tube furnaceis a topic worth studying.In order to efficiently and accurately collect tube furnace flame image, the paperdesign of flame image acquisition system. It’s constituted of the image transmissionoptical fiber, CCD camera, the cooling system, image acquisition card, videodistributor, etc. Not only according to the requirement of the new abnormalcombustion identification and the characteristic of the various hardware, select areasonable hardware; But also according to the cause of the abnormal combustionflame appears, judge the abnormal flame appeared regularly in the heating pipewelding, thus reasonable installation position of flame detector,extract effectivelyflame video image.Based on the theory of noise, analyzes the source of image noise,furnace flamecombustion process is accompanied by a large number of dynamic random andisolated noise, the noise signal must be eliminated. In this paper, the median filter caneffectively eliminate the image noise signal. Simultaneously aims at the problem thatin flame image the contrast of the flame goal and the background is not strong,proposed the use gradation linear substitution strengthens the flame goal and thebackground contrast gradient. And effectively remove the edge burr and tiny holesin the flame targets using morphology. Through the above image preprocessingoperations, effectively improve the flame target extraction accuracy and processingspeed.Due to the complexity of the background in the hearth and the influence offlame crosstalk,using the traditional segmentation algorithm to extract the flame target can’t get a satisfactory effect. Based on the relation between the flame video frames,by analysis of existing image segmentation techniques, we propose to use symmetricdifference algorithm, effectively extract the flame goal. By analyzing thecharacteristics of the flame abnormality,put forward an improved center distanceclustering algorithm to identify effectively abnormal flame target. The experimentproved that this topic selection algorithms in the tube furnace abnormal new flameburns recognition has better accuracy and reliability.