| There is a widespread phenomenon of high temperature of the earth's surface.The existence of high temperature phenomenon is closely related to human social life.Its research has important significance to people survival.High temperature target is usually the temperature is above 500 K and the temperature is significantly higher than normal temperature features around special plant-based.In addition to the oil well torch,heating and power generation factory chimneys,and other high temperture normal target,more high temperature targets hidden danger existing in People's Daily life.Such as coke making by indigenous methods,the environmental pollution caused by the fire,fire and other volcanic eruptions will cause huge economic loss and may even cause irreparable loss of life.So to find the fire in time,preventing the happening of the loss is very necessary.By means of remote sensing on target identification,high precision positioning and its temperature is calculated.That can real-time dynamic monitoring the occurrence of some high temperature disaster and save a lot of manpower material resources.Using remote sensing technology to monitor temperature target also facilitates rapid decisions to find out the feasible solution.In the paper,we use short wave infrared temperature inverion is achieved with the method of physical modeling.That model is a lot difficule than satistical model.But it have better expansibility and avoid the limitations of the statistical model.Following the physical laws of remote sensing system to establish the causal relationship to abstract reality can build a more realistic model.And subsequent validation of the model is the focus of this aricel research content.Satellite transit from the wild synchronous test to rule out a lot of restrictions can be repeated many times of indoor simulation experiment.It is the important link of the high temperature target of the short wave infrared temperature inversion model validation.Of the commonly used surface temperature inverion in the paper,we did not use the classic band thermal infrared ban(TIRS,8~14?m)temperature inversion,instead more sensitive to high temperature target of short wave infrared(SWIR,1.3~2.5?m)of high temperature target recognition of inversion.Thermal infrared temperature inversion contain the whole pixel's temperature(In most cases,the high temperature target failed to fill the whole pixel.)Short wave infrared inversion can obtain high temperature target itself temperature in the mixed pixel.Usually we obtain the spatial resolution of remote sensing data in short wave infrared wavelengths are much better than that of thermal infrared band.And short wave infrared wavelengths are less likely to reach saturation than thermal infrared band,namely in the broader threshold can get more accurate results.According to the blackbody radiation law,with the increasing of temperature,peak wavelength of blackbody radiation energy is moving in the direction of short wave infrared.Therefore,short wave infrared wavelengths for high temperature targets(especially small target high temperature object recognition)inversion from thermal infrared band has more significant advantages.The temperature inversion of high temperature target,first of all need to identify high temperature pixel.First,statistics on the high temperature target is different from most of the normal background features.And they are a few of the abnormal points.Next,spectral characteristic on the spectrum characteristics of high temperature target is a significant difference between interesting feature in the background.Thus from the vision,the color on the remote sensing image can also be used with the surrounding normal temperature significantly distinguish feature category.In order to better understand the characteristics of high temperature target,effective recognition extracted form a lot normal temperature background pixel that small number of high temperature pixel.Study a large number of different types of ground objects spectrum characteristics.This paper respectively expounds two ways to obtain all kind of features of the spectrum.One kind is various features of the measured spectra,and the other kind is selection of the area of interest on the remoter sensing image for image spectrum.This paper mainly introduces the fire(high temperature)to construct and other identification method for two kinds.They are normalized differential Fire index method(NDFI),analysis of variance(separability measure),factor analysis and correspondence analysis.Othe method is multivariate markov distance truncation method,markov distance discriminant method and Fisher of two types of discrimination.This are offshore oil wells in torch,making coke,oil tank explosion,fire and volcanic eruptions.In the paper,we focus on three parts.Respectively high temperature and shortwave infrared temperature inversion model,high temperature target recognition method research and outdoor and indoor test experiment of the model.The main research results are as follows:1.For the study of special features of high temperature target,short wave infrared has higher sensitivity than thermal infrared.On the saturation,short-wave infrared with a larger capacity limit,so short wave infrared are favourable band for high temperature targets research.High temperature target short wave infrared model involved in several key parameters such as the reflection of high-temperature target launch,normal temperature object,atmospheric transmittance,high temperature target pixel area percentage and the goal of high temperature.The launch of high temperature target,normal temperature features of reflection,the estimation of atmospheric transmittance are key content of the study.Their accuracy is directly related to high temperature target short wave infrared temperature on the accuracy of the inversion results.High temperature targets short wave infrared model research can obtain high temperature information quickly and efficiently.The research can provide solutions and theoretical reference for disasters.The identification results of multivariate markov distance truncation method,markov distance discriminant method,factor analysis and correspondence analysis are 82%,79%,85% and 92%.2.This paper respectively by variance analysis,factor analysis,correspondence analysis and multivariate markov distance truncation method,markov distance discriminant method,respectively,for a small target(making)and large(fire)was studied to identify target.Three identification methods are based on the spectrum characteristics differences between high temperature goal and other various types of ground objects.Two recognition method are based on high temperature goal and all kinds of other object statistical characteristics.A lot of the statistical characteristics of spectral are numerical collected and summarized in the research.We use remote sensing technology for all kinds of high temperature in the extract target recognition.According to different research areas,the precision of each method are different.For the high temperature target in other temperature characteristics of the object can be more reasonable and effective method to choose the suitable for high temperature target recognition research.3.The high temperature short wave infrared temperature inversion model is reasonable and effective.Based on the research of the field and laboratory experiments we verify it.By control of high temperature target,the background(normal temperature terrain background reflectivity),temperature of high temperature targets,the area percentage of the pixel(by adjusting the probe to control the indoor experiments)these different variables to enrich the verification experiment.Thus we can validated the effectiveness of the research more comprehensive for the high temperature short wave infrared temperature inversion model.Field validation experiments affected by external conditions(environment)is bigger,so the indoor verification experiment research carried out can guarantee the verification experiment. |
PDF Full Text Request