ClassificationSlide 2Slide 3Landsat TM BandsSpectral Resolution of Landsat TMFeature Space (2 bands)Feature SpaceSlide 8Unsupervised ClassificationComputer Algorithm Finds ClustersSlide 11Problems with Unsupervised ClassificationSupervised ClassficationSupervised ClassificationProblems with Supervised ClassificationThe Result: A Classified ImageCrop ClassificationClassificationClassification•Key remote sensing processing technique•Placing pixels into thematic categories that correspond to land cover types–e.g. forest, crops, water, urban, etc.•Basis for classification are the spectral signatures of landcover typesLandsat TM BandsSpectral Resolution of Landsat TMTM Band: 1 2 3 4 5These bands provide a coarse summary of spectral signatures.(NIR) (MIR)Feature Space(2 bands)Red BV’sNIR BV’s00 255255vFeature SpacePixels in the same classes clusterwatersoilagriforestRed BV’sInfrared BV’s00 255255Classification•Two essential types of classification–Unsupervised •Unsupervised by humans•Computer algorithm-based–Supervised•Supervised by human(s)•Human-trained•Algorithms also usedUnsupervised Classification•Classes are created based on the locations in feature space of the pixel dataRed BV’sInfrared BV’s00 255255vComputer Algorithm Finds ClustersRed BV’sInfrared BV’s00 255255vUnsupervised ClassificationUnsupervised Classification•Attribution phase – performed by humanwaterSoilagriforestRed BV’sInfrared BV’s00 255255Problems with Unsupervised ClassificationRed BV’sInfrared BV’s00 255255vThe computer may consider these 2 clusters (forest and agri) as one clusterThe computer may consider this cluster to be 2 clustersSupervised Classfication•We “train” the computer program using ground truth dataConiferous treesDeciduous treesSupervised ClassificationRed BV’sInfrared BV’s00 255255vSample pixelsOther pixelsProblems with Supervised ClassificationRed BV’sInfrared BV’s00 255255vforestagriwaterSoilWhat’s this?vThe Result: A Classified ImageGauja National Park May 20020 9 18 27 364.5KilometersLegend1234567Coniferous forestshrubshardwood
View Full Document