Hydrocarbon Information Extraction Based On Spectral Unmixing In Hyperspectral Remote Sensing

Since the 80’s of The last century, with the rapid development of remote sensing technology, by means of the delineation of oil and gas exploration target area for oil through remote sensing technology of extracting hydrocarbon micro-seepage information, has become a new trend in the development of the modern oil and gas resource exploration. As an emerging technology of this modern society, hyperspectral remote sensing technology has been successfully applied to in many important areas, such as geological mapping, environmental monitoring, vegetation coverage, marine monitoring, it covers satellite observation, aerial observation and observation of earth based on the detection of advanced technology, the development of the image, the spectral information it develops researching about synthetical data processing, information extraction and analysis, the quantitative application of, multi-disciplinary, multi field. Due to the limitation of spatial resolution about the remote sensing imaging instrument as well as the influence of land cover types in nature complex, it result in mixed pixels in the remote sensing image, the image pixel level precision is affected, the classification of remote sensing and measurement accuracy can not meet the requirements. This has a direct impact on the accurate analysis and application of hyperspectral data. What we are interested in is, in the sub-pixel or is in the weak state. Therefore, hysperctral unmixing of remote sensing, has become an urgent problem to be solved, and it is also a difficult topic of image target detection in hyperspectral remote sensing.According to the prior information, the hyperspectral target detection algorithms can be divided into supervised classification and unsupervised classification of target detection, this paper summarizes five kinds of target detection algorithm, orthogonal subspace projection algorithm for supervised classification of the study, which puts forward the goal of constructing the vector to the interested pixel subspace projection algorithm for spectral unmixing, and we compared with the orthogonal subspace projection algorithm for target of interest, and the result is consistent with the corresponding pixel vector abundance. Orthogonal subspace projection focuses on a single target abundance information, other information as background information is compressed into zero space; and direct projection algorithm focuses on the overall goal, calculates the abundance of all targets, characterizes n the pixel. Finally, based on the study of hydrocarbon micro seepage principle and anomaly information of remote sensing mechanism, this thesis combines the linear spectral mixing model solution of hyperspectral remote sensing target detection method, adopt the pure pixel index of endmember extraction method to analysis the possible information Airborne Hyperspectral remote sensing image of hydrocarbon micro seepage, based on the characteristics of space projection of mixed solution to identify pixels, this paper extract the hydrocarbon micro seepage weak information, apply it to the engineering application, and apply the abnormal erosion caused by the coverage area of hydrocarbon micro seepage of oil and gas in Jimusar area of information extraction, it achieved good results. It provides effective data feature space projection of the hydrocarbon weak information extraction for the previous work of field exploration, which offers a reference for the oil and gas delineation of target areas, it exert an positive effect the academic research and engineering application.