The Algorithms Research On Data Processing Of Infrared Imaging Spectrometer

Infrared imaging spectrometer can gain the spatial distribution information and the spectrum information of the measured target at the same time.It has the unique abilities of getting information and identifying characteristics,which make it can be widely used in many fields.This thesis performed image processing of the data measured by the jir5001 on-site multiband infrared imaging spectrometer.We mainly did the research on dimension reduction,classification,endmember extraction and unmixing of the mixed pixel for the image.First,we studied three dimension reduction algorithms of the infrared imaging spectrometer image,such as principal component analysis(PCA),the adaptive band selection method and wavelength selection method based on spectral angle mapping.Then we studied a classification method based on spatial continuity to classify infrared imaging spectrometer image.Making full use of the spectral information and the spatial distribution information of the infrared imaging spectrometer image to deal with the original image for block processing,and then classify the new data we get.We simulated the algorithm and extract the target material of the original image data and the image data after block processing.The result shows that after block processing based on spectral feature and spatial continuity classifying the infrared imaging spectrometer image will improve its accuracy.Finally we studied the endmember extraction algorithm for infrared imaging spectrometer image,such as PPI algorithm,N-FINDR algorithm.After studying and analyzing those algorithms,we chose N-FINDR based on PPI algorithm to do endmember extraction for the infrared imaging spectrometer image.The result shows that the algorithm can effectively extract the endmember and reduce the calculation cost.And the unmixing algorithm based on the volume of the simplex is studied.This algorithm is used to unmix the infrared imaging spectrometer image.The abundance figure of each endmember in the image are calculated.Since this algorithm uses the relationship between content and volume ratio of pixels and endmembers,it has lower time complexity and reduce the calculation cost.