ENVI Tutorial
25
pages
English
Documents
Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres
25
pages
English
Ebook
Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres
ENVI Tutorial: ClassificationMethodsClassification Methods 2Files Used in this Tutorial 2Examining a Landsat TM Color Image 3Reviewing Image Colors 3Using the Cursor Location/Value 4Examining Spectral Plots 4Exploring Unsupervised Classification Methods 6Applying K-Means Classification 6Applying ISODATA Classification 7Exploring Supervised Classification Methods 9Selecting Training Sets Using Regions of Interest (ROI) 9Applying Parallelepiped Classification 9Applying Maximum Likelihood Classification 10Applying Minimum Distance Classification 10Applying Mahalanobis Distance Classification 11Collecting Endmember Spectra 11Applying Binary Encoding Classification 12Exploring Spectral Classification Methods 14Exploring Rule Images 15Post Classification Processing 17Extracting Class Statistics 17Generating a Confusion Matrix 18Clumping and Sieving 19Combining Classes 20Overlaying Classes 20Editing Class Colors 22Working with Interactive Classification Overlays 23Overlaying Vector Layers 24Converting a Classification to a Vector 24Adding Classification Keys Using Annotation 251ENVI Tutorial: Classification MethodsClassification MethodsThis tutorial provides an introduction to classification procedures using Landsat TM data from CañonCity, Colorado. Results of both unsupervised and supervised classifications are examined and postclassification processing including clump, sieve, combine classes, and accuracy assessment arediscussed.Files ...
Voir
ENVI Tutorial: Classification Methods
Classification Methods Files Used in this Tutorial Examining a Landsat TM Color Image Reviewing Image Colors Using the Cursor Location/Value Examining Spectral Plots Exploring Unsupervised Classification Methods Applying K-Means Classification Applying ISODATA Classification Exploring Supervised Classification Methods Selecting Training Sets Using Regions of Interest (ROI) Applying Parallelepiped Classification Applying Maximum Likelihood Classification Applying Minimum Distance Classification Applying Mahalanobis Distance Classification Collecting Endmember Spectra Applying Binary Encoding Classification Exploring Spectral Classification Methods Exploring Rule Images Post Classification Processing Extracting Class Statistics Generating a Confusion Matrix Clumping and Sieving Combining Classes Overlaying Classes Editing Class Colors Working with Interactive Classification Overlays Overlaying Vector Layers Converting a Classification to a Vector Adding Classification Keys Using Annotation
1
2 2 3 3 4 4 6 6 7 9 9 9 10 10 11 11 12 14 15 17 17 18 19 20 20 22 23 24 24 25
ENVI Tutorial: Classification Methods
Classification Methods This tutorial provides an introduction to classification procedures using Landsat TM data from Cañon City, Colorado. Results of both unsupervised and supervised classifications are examined and post classification processing including clump, sieve, combine classes, and accuracy assessment are discussed. Files Used in this Tutorial _ ENVI Resource DVD:Data\can tm File Description can_tmr.imgCañon City, Colorado TM reflectance image _ can tmr.hdrENVI header for above _ can km.imgK-means classification _ can km.hdrENVI header for above _ can iso.imgISODATA classification _ can iso.hdrENVI header for above classes.roiRegions of interest (ROI) for supervised classification can_pcls.imgParallelepiped classification _ can pcls.hdrENVI header for above _ mgBinary encoding can bin.iresult _ can bin.hdrENVI header for above _ can sam.imgSAM classification result _ can sam.hdrENVI header for above can_rul.imgRule image for SAM classification _ can rul.hdrENVI header for above _ge can sv.imgSieved ima _ can sv.hdrENVI header for above can_clmp.imgClump of sieved image _ can clmp.hdrENVI header for above _g can comb.img eCombined classes ima _ can comb.hdrENVI header for above can_ovr.imgClasses overlain on gray scale image _ can ovr.hdrENVI header for above _ can v1.evfVector layer generated from class #1 _ can v2.evfVector layer generated from class #2
2
ENVI Tutorial: Classification Methods
Examining a Landsat TM Color Image This portion of the exercise will familiarize you with the spectral characteristics of the Landsat TM data of Cañon City, Colorado, USA. Color composite images will be used as the first step in locating and identifying unique areas for use as training sets in classification. Before attempting to start the program, ensure that ENVI is properly installed as described in the Installation Guide that shipped with your software. 1. From the ENVI main menu bar, selectFile > Open Image File. _ _ 2. Navigate to theData\can tmdirectory, select the filecan tmr.imgfrom the list, and click Open. The Available Bands List appears on your screen. 3. Click on theRGB ColorList. Red, Green, and Blue fieldsradio button in the Available Bands appear in the middle of the dialog. 4. SelectBand 4,Band 3, andBand 2sequentially from the list of bands at the top of the dialog by clicking on the band names. The band names are automatically entered in the Red, Green, and Blue fields. 5. ClickLoad RGBto load the image into ENVI. 6. Examine the image in the display group. Reviewing Image Colors The color image displayed below can be used as a guide to classification. This image is the equivalent of a false color infrared photograph. Even in a simple three-band image, it’s easy to see that there are areas that have similar spectral characteristics. Bright red areas on the image represent high infrared reflectance, usually corresponding to healthy vegetation, either under cultivation, or along rivers. Slightly darker red areas typically represent native vegetation, in this case in slightly more rugged terrain, primarily corresponding to coniferous trees. Several distinct geologic and urbanization classes are also readily apparent.
3
ENVI Tutorial: Classification Methods
Using the Cursor Location/Value Use ENVI’s Cursor Location/Value option to preview image values in the displayed spectral bands. 1. From the Display group menu bar, selectTools > Cursor Location/Value. Alternatively, double-click the left mouse button in the Image window to toggle the Cursor Location/Value dialog on and off.
2. Move the cursor around the image and examine the data values in the dialog for specific locations. Also note the relation between image color and data value. 3. From the Cursor Location/Value dialog, selectFiles > Cancel. Examining Spectral Plots Use ENVI’s integrated spectral profiling capabilities to examine the spectral characteristics of the data. 1. From the Display group menu bar, selectTools > Profiles > Z Profile (Spectrum)to begin extracting spectral profiles. 2. Examine the spectra for areas that you previewed above using color images and the Cursor/Location Value dialog by clicking the left mouse button in any of the display group
4
3.
ENVI Tutorial: Classification Methods
windows. Note the relations between image color and spectral shape. Pay attention to the location of the image bands in the spectral profile, marked by the red, green, and blue bars in the plot.
From the Spectral Profile dialog menu bar, selectFile > Cancel.
5
ENVI Tutorial: Classification Methods
Exploring Unsupervised Classification Methods Unsupervised classification can be used to cluster pixels in a dataset based on statistics only, without any user-defined training classes. The available unsupervised classification techniques are K-Means and ISODATA. Applying K-Means Classification K-Means unsupervised classification calculates initial class means evenly distributed in the data space, then iteratively clusters the pixels into the nearest class using a minimum-distance technique. Each iteration recalculates class means and reclassifies pixels with respect to the new means. All pixels are classified to the nearest class unless a standard deviation or distance threshold is specified, in which case some pixels may be unclassified if they do not meet the selected criteria. This process continues until the number of pixels in each class changes by less than the selected pixel change threshold or the maximum number of iterations is reached. 1. From the ENVI main menu bar, selectClassification > Unsupervised > K-Meansor review the _ _ pre-calculated results of classifying the image by opening thecan km.imgfile in thecan tm directory. 2. Select thecan tmr.imgfile and clickOK. The K-Means Parameters dialog appears. _
3. Accept the default values, select theyoremMradio button, and clickOK. The new band is loaded into the Available Bands List. 4. From the Available Bands List, click theDisplay #1button and selectNew Display. 5. From the Available Bands List, select thesK-anMeband and clickLoad Band. 6. From the Display group menu bar, selectTools > Link > Link Displaysthen clickOKto link the images.
6
ENVI Tutorial: Classification Methods
7. Compare the K-Means classification result to the color-composite image using the dynamic overlay feature in ENVI (click using the left mouse button in the Image window). 8. From the Display group menu bar, selectTools > Link > Unlink Displayto remove the link and turn off the dynamic overlay feature. 9. If desired, experiment with different numbers of classes, change thresholds, standard deviations, and maximum distance error values to determine their effect on the classification. Applying ISODATA Classification ISODATA unsupervised classification calculates class means evenly distributed in the data space then iteratively clusters the remaining pixels using minimum distance techniques. Each iteration recalculates means and reclassifies pixels with respect to the new means. This process continues until the number of pixels in each class changes by less than the selected pixel change threshold or the maximum number of iterations is reached. 1. From the ENVI main menu bar, selectClassification > Unsupervised > IsoData, or review the _ _ pre-calculated results of classifying the image by opening thecan iso.imgfile in thecan tm.yidertcro 2. Select thecan tmr.imgfile and clickOK. The ISODATA Parameters dialog appears. _ 3. Accept the default values, select theMmeroyradio button, and clickOK. The new band is loaded into the Available Bands List.
4. From the Available Bands List, click theDisplay #2button and selectNew Display. 5. Select theSIATODAband and clickLoad Band. 6. From the Display group menu bar, selectTools > Link > Link Displays. The Link Displays dialog appears.
7
ENVI Tutorial: Classification Methods
7. 8. 9. 10. 11. 12. 13.
Click theDisplay #2toggle button to selectNo, and click theDisplay #3toggle button to select Yes. ClickOKto link the images. Compare the ISODATA classification result to the color-composite image using the dynamic overlay feature in ENVI (click using the left mouse button in the Image window). From the Display group menu bar, selectTools > Unlink Displays. From the Display group menu bar, selectTools > Link > Link Displays. The Link Displays dialog appears. Click theDisplay #1toggle button to selectNo, and ensure that theDisplay #2andDisplay #3 toggle buttons say Yes. Click OK to link and compare the K-means and ISODATA images. If desired, experiment with different numbers of classes, change thresholds, standard deviations, maximum distance error, and class pixel characteristic values to determine their effect on the classification. From the Display group menu bar on the K-Means Image window, selectFile > Cancelto close the display group. Close the ISODATA display group using the same technique.
8
ENVI Tutorial: Classification Methods
Exploring Supervised Classification Methods Supervised classification can be used to cluster pixels in a dataset into classes corresponding to user-defined training classes. This classification type requires that you select training areas for use as the basis for classification. Various comparison methods are then used to determine if a specific pixel qualifies as a class member. ENVI provides a broad range of different classification methods, including Parallelepiped, Minimum Distance, Mahalanobis Distance, Maximum Likelihood, Spectral Angle Mapper, Binary Encoding, and Neural Net. In this tutorial, you will experiment with two methods for selecting training areas, also known as regions of interest (ROIs). Selecting Training Sets Using Regions of Interest (ROI) As described in the tutorial,An Introduction to ENVIand summarized here, ENVI lets you define regions of interest (ROIs) typically used to extract statistics for classification, masking, and other operations. For the purposes of this exercise, you can either use predefined ROIs, or create your own. In this exercise, you will restore predefined ROIs. 1. From the #1 Display group menu bar, selectTools > Region of Interest > ROI Tool. The ROI Tool dialog appears. 2. From the ROI Tool dialog menu bar, selectFile > Restore ROIs. The Enter ROI Filenames dialog appears. 3. Select thecalssse.roifile and clickOpen. ClickOK. The ROIs appear in the Image window. Applying Parallelepiped Classification Parallelepiped classification uses a simple decision rule to classify multispectral data. The decision boundaries form an n-dimensional parallelepiped classification in the image data space. The dimensions of the parallelepiped classification are defined based upon a standard deviation threshold from the mean of each selected class. If a pixel value lies above the low threshold and below the high threshold for all n bands being classified, it is assigned to that class. If the pixel value falls in multiple classes, ENVI assigns the pixel to the last class matched. Areas that do not fall within any of the parallelepiped classifications are designated as unclassified. 1. From the ENVI main menu bar, selectClassification > Supervised > Parallelepiped, or review _ the pre-calculated results of classifying the image by opening thecan pcls.imgfile in the _ can tmdiro.yertc _ 2. Select thecan tmr.imgfile and clickOK. The Parallelepiped Parameters dialog appears. 3. Click theSelect All Itemsbutton to select the ROIs. 4. Select to output the result toyoremMusing the radio button provided. 5. Click theOutput Rule Imagestoggle button to selectNo, then clickOK. The new band is loaded into the Available Bands List.
9
ENVI Tutorial: Classification Methods
6. From the Available Bands List, click theDisplay #1button and selectNew Display. 7. Select thelarPlealband and clickLoad Band. 8. From the Display group menu bar, selectTools > Link > Link Displaysand clickOKin the dialog to link the images. 9. Use image linking and dynamic overlay to compare this classification to the color composite image. Applying Maximum Likelihood Classification Maximum likelihood classification assumes that the statistics for each class in each band are normally distributed and calculates the probability that a given pixel belongs to a specific class. Unless a probability threshold is selected, all pixels are classified. Each pixel is assigned to the class that has the highest probability (i.e., the maximum likelihood). 1. Using the steps above as a guide, perform a Maximum Likelihood classification. 2. Try using the default parameters and various probability thresholds. 3. Use image linking and dynamic overlay to compare this classification to the color composite image and previous unsupervised and supervised classifications. Applying Minimum Distance Classification The minimum distance classification uses the mean vectors of each ROI and calculates the Euclidean distance from each unknown pixel to the mean vector for each class. All pixels are classified to the closest ROI class unless the user specifies standard deviation or distance thresholds, in which case some pixels may be unclassified if they do not meet the selected criteria. 1. Using the steps above as a guide, perform a Minimum Distance classification. 2. Try using the default parameters and various standard deviations and maximum distance errors.
10
ENVI Tutorial: Classification Methods
3. Use image linking and dynamic overlay to compare this classification to the color composite image and previous unsupervised and supervised classifications. Applying Mahalanobis Distance Classification The Mahalanobis Distance classification is a direction sensitive distance classifier that uses statistics for each class. It is similar to the Maximum Likelihood classification but assumes all class covariances are equal and therefore is a faster method. All pixels are classified to the closest ROI class unless you specify a distance threshold, in which case some pixels may be unclassified if they do not meet the threshold. 1. Using the steps above as a guide, perform a Mahalanobis Distance classification. 2. Try using the default parameters and various maximum distance errors. 3. Use image linking and dynamic overlay to compare this classification to the color composite image and previous unsupervised and supervised classifications. 4. When you are finished, close all classification display groups. Collecting Endmember Spectra The Endmember Collection:Parallel dialog is a standardized means of collecting spectra for supervised classification from ASCII files, ROIs, spectral libraries, and statistics files. 1. From the ENVI main menu bar, selectClassification > Endmember Collection. The Classification Input File dialog appears. 2. Select thecan tmr.imgfile and clickOK. _
11
© 2010-2024 YouScribe
