Feature Extraction Of Soot Image

As air pollution becomes more and more serious in recent years,PM2.5 keep up increasingly,which mainly comes from three sources: vehicle emissions,coal and dust.With the improvement of people's living standard,cars have become an important means of transportation.As a result,the diesel engine has been widely used because the diesel engine has the advantages of high thermal efficiency,fuel economy,long life,superior power,and so on.However,as the diesel fuel has a higher percentage of carbon content relative to gasoline,diesel particulate matter(PM)emissions are much higher than the gasoline engine emissions,and it is urgent to reduce diesel emissions.Consequently,in order to reduce diesel particulate matter pollution,it is necessary to study the formation mechanism of diesel engine particles.Through the imaging of diesel particulate soot by high-resolution transmission electron microscopy,it provides a direct information on lattice fringe structure of the soot in the atomic level,and effectively promotes the research of particle formation mechanism.At the same time,in order to quantify the morphology of soot,the method of feature extraction and analysis of high-resolution TEM image must be applied,that is the study in this thesis that calls Research Feature Extraction of Fringe in HRTEM Soot Image.In this thesis,the feature extraction of fringe in soot images is divided into three parts: image processing method of soot,fringe correction based on fuzzy reasoning,and extraction method of fringe structure.The research of this thesis has the following advantages compared with the existing study of fringe in soot:(1)By combining the use of Tophat and DFT bandpass filtering to enhance the fringe in the soot image,Tophat has the suppression of noise and optimize bandpass band of DFT.Meanwhile,the enhanced filter is applied to enhance bandpass band of DFT,and achieved good fringe enhancement effect.(2)Based on the fuzzy reasoning,the fringe is modified,and the fuzzy reasoning rules are designed for the interruption of Y-type or T-type microcrystals and the connection of all the two-endpoint fringe.The reliability of the fringe structure is reasonably guaranteed.In addition,fuzzy reasoning is easy to introduce a priori knowledge,increase the preference for the problem area,and ensure the model is simple and effective.(3)The accuracy of the algorithm is verified by the inter-layer spacing in the HRTEM image of the standard carbon.The maximum error of the interlayer interval is less than 0.018 nm and the average error is about 0.0072 nm,which satisfies the high precision Claim.(4)By maximizing the automation of the algorithm and software,the developed image contrast plug-in greatly facilitate the algorithm design process and the use of the parameters of the adjustment.