Using GIS Introduction to GIS Lecture 12 More

------Using GIS-- Introduction to GIS Lecture 12: More Remote Sensing 1. Major Types of Satellite Imagery 2. Remote Sensing Image Interpretation and Classification By Austin Troy © 2003 University of Vermont All materials by Austin Troy © 2003

------Using GIS-- Introduction to GIS Part 1: Major types of satellite imagery All materials by Austin Troy © 2003

Introduction to GIS Major satellite imagery products • SPOT • Landsat TM • Landsat MSS • IKONOS All materials by Austin Troy © 2003

Introduction to GIS SPOT • Launched by France • Stands for Satellite Pour l'Observation de la Terre • Operated by the French Space Agency, Centre National d'Etudes Spatiales (CNES). All materials by Austin Troy © 2003

Introduction to GIS SPOT • SPOT 1 launched 1986, decommissioned and the reactivated in 1997 • SPOT 2 launched 1990, still going • SPOT 3 launched 1993 and stopped functioning 1996 • SPOT 4 launched in 1998, still going • SPOT 5 scheduled for April 2002 All materials by Austin Troy © 2003

Introduction to GIS SPOT • SPOT satellites are in sun-synchronous orbit • The satellite passes over the same part of the Earth at roughly the same local time each day • Its “inclination” is about 8 degrees off of polar orbit • The fact that the earth is not perfect sphere makes the orbital plane rotate slowly around the earth (this would not happen if it were perfectly polar) All materials by Austin Troy © 2003

Introduction to GIS SPOT • The slow motion of that orbital plane matches the latitudinal motion of the sun in the sky over the year • Maintains similar sun angles along its ground trace for all orbits • That means that the area the sun flies over always get the same sunlight angle, which gives constant lighting All materials by Austin Troy © 2003 Source: http: //hdsn. eoc. nasda. go. jp/experience/rm_kiso/sat ellit_type_orbit_e. html

Introduction to GIS SPOT This is for LANDSAT, but the idea is the same for SPOT Source: http: //ltpwww. gsfc. nasa. gov/IAS/handbook/handbo ok_htmls/chapter 6. html All materials by Austin Troy © 2003

Introduction to GIS SPOT • Each SPOT satellite carries two HRV (high-resolution visible) sensors, constructed with multilinear array detectors, or “pushbroom scanners”, also known as “along track scanners” • These record multispectral image data along a wide swath Source: http: //www. sci-ctr. edu. sg/ssc/publication/remotesense/spot. htm All materials by Austin Troy © 2003

Introduction to GIS SPOT • Pushbroom uses a “linear array” of detectors, so it senses single column at a time, and uses forward motion to generate second dimension • GSD (ground sampled distance), or resolution is set by sampling interval. Normally results in just-touching square pixels making up the image • Each spectral band of sensing requires its own array. • Pushbroom scanners generally have higher radiometric resolution because they have longer “dwell time” than acrosstrack scanners, which move laterally across landscape as also move forward All materials by Austin Troy © 2003

Introduction to GIS SPOT • The position of each HRV unit can be changed by ground control to observe a region of interest that is at an oblique angle to the satellite—up to ± 27º relative to the vertical. • Off-nadir viewing allows for acquisition of stereoscopic imagery (because of the parallax created) and provides a shorter revisit interval of 1 to 3 days. Source: http: //www. sci-ctr. edu. sg/ssc/publication/remotesense/spot. htm All materials by Austin Troy © 2003

Introduction to GIS SPOT • Oblique viewing capacity allows it to image any area within a 900 kilometer swath; can be used to increase the viewing frequency for a given point during a given cycle. The frequency varies with latitude: at the equator, a given area can be imaged 7 times during the same 26 -day orbital cycle. At latitude 45 degrees, a given area can be imaged 11 times during the orbital cycle, i. e. 157 times yearly and an average of 2. 4 days, with an interval ranging from a maximum of 4 days to a minimum of 1 day. • Any point on 95% of the earth may be imaged any day by one of the three satellites. Source: http: //www. spot. com/home/system/introsat/acquisi/ welcome. htm All materials by Austin Troy © 2003

Introduction to GIS SPOT • Two modes: panchromatic and multispectral • Panchromatic: single spectral band, corresponding to the visible part of the EM spectrum without the blue, from 0. 51 to 0. 73 µm. Single channel imaging mode, so yields black and white images. Resolution is 10 m. Pixels per line is 6000. Good for fine geometrical detail. • Multispectral mode: three spectral bands are XS 1 covering 0. 50 to 0. 59 µm (green), XS 2 covering 0. 61 to 0. 68 µ m (red) and XS 3 covering 0. 79 to 0. 89 µm (near infrared). Resolution is 20 m. Pixels per line is 3000. All materials by Austin Troy © 2003

Introduction to GIS SPOT • Some examples: mosaic false color tiles of Australia All materials by Austin Troy © 2003

Introduction to GIS SPOT All materials by Austin Troy © 2003

Introduction to GIS SPOT • SPOT can be purchased online using a browser to select your area and product http: //www. spot. com/HOME/PROSER/USASelect_On. Line/usa_select_on-line. htm All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • first started by NASA in 1972 but later turned over to NOAA • Since 1984 satellite operation and data handling are managed by a commercial company EOSAT Source: http: //www. sci-ctr. edu. sg/ssc/publication/remotesense/landsat. htm All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • LANDSAT-1 launched 1972 and lasted until 1978. • LANDSAT-2 launched 1975 • Three more satellites were launched in 1978, 1982, and 1984 (LANDSAT-3, 4, and 5 respectively). • LANDSAT-6 was launched on October 1993 but the satellite failed to obtain orbit. • LANDSAT-7 launched in 1999 • Only 7 and 5 are still working All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • Like SPOT, LANDSAT is sun-synchronous, and is about 8 degrees off a polar orbit • Its repeat cycle is about 16 days and always crosses equator at around 10 AM. • Orbit takes about 99 minutes (14. 5 per day) • Distance between ground tracks of consecutive orbits is 2752 km at equator because of the earth’s rotation • By following earth’s rotation with each pass, it can keep crossing the equator at the same time All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • Swath is 183 km wide, although that includes overlap, since data frame is 170 km • 233 orbits, for each 16 day cycle Source: http: //eosims. cr. usgs. gov: 5725/DATASET_DOCS/landsat 7_dataset. html All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • Scenes are then indexed by the path and a row Source: http: //eosims. cr. usgs. gov: 5725/DATASET_DOCS/landsat 7_dataset. html All materials by Austin Troy © 2003

Introduction to GIS LANDSAT • LANDSAT 4 and 5 had two types of sensors, MSS (multi-spectral scanner) and TM (thematic mapper): • MSS: Started on LANDSAT 1, terminated in late 1992. 80 m resolution with four spectral bands from the visible green to the near-infrared (IR) wavelengths. Only Landsat 3’s MSS sensor had a fifth band in thermal-IR. All materials by Austin Troy © 2003

Introduction to GIS LANDSAT 4 and 5 MSS: • TM: * Mid infra red * * All materials by Austin Troy © 2003

Introduction to GIS LANDSAT MSS • MSS has a square instantaneous field of view (IFOV), with an 11. 56 ° field of view. Source: http: //edcwww. cr. usgs. gov/glis/hyper/guide/landsat All materials by Austin Troy © 2003

Introduction to GIS LANDSAT MSS • This is a “whiskbroom” rather than “pushbroom” scanner. AKA Across track scanning • Satellite motion provides one axis of the image and other axis provided for by oscillating mirror • Has poor radiometric resolution- only 6 bit, or 64 values All materials by Austin Troy © 2003

Introduction to GIS LANDSAT TM • Thematic Mapper: more bands, better spatial and radiometric resolution(256 DNs instead of 64) • Both resolution improvements, plus the fact that the green and red bands are narrower make it better for vegetation discrimination than MSS; also near IR in TM is narrower and centered in a region that is highly sensitive to plant vigor. All materials by Austin Troy © 2003

Introduction to GIS LANDSAT TM: applications Band Nominal Spectral location applications 1 Blue Water body penetration, soil-water discrimination, forest type mapping, cultural feature ID 2 Green reflectance peak of veg, for veg ID and assessment of vigor, cultural feature ID 3 Red Chlorophyll absorption region, plant species differentiation, cultural feature ID 4 Near infra red Veg types, vigor and biomass content, dilineating water bodies, soil moisture assessment 5 mid infra red (1. 551. 75 mm) Veg moisture, soil moisture, diff of soil from clouds 6 Thermal infra red Veg stress analysis, soil moisture, thermal mapping 7 mid infra red(2. 08 -2. 35 Discriminating mineral and rock types, veg moisture mm) All materials by Austin Troy © 2003

Introduction to GIS LANDSAT TM • An example: August 14, 1999 (left) and October 17, 1999 (right) images of the Salt Lake City area • differences in color due to growing season All materials by Austin Troy © 2003

Introduction to GIS LANDSAT 7 • Uses a new sensor called Enhanced Thematic Mapper Plus (ETM+) • Stresses continuity with LANDSAT 4 and 5 in that uses similar orbit and repeat patterns, as well as a similar 185 km swath width for imaging Source: http: //ltpwww. gsfc. nasa. gov/IAS/handbook_htmls/chapter 2. html • Check out the movie Full info at http: //ltpwww. gsfc. nasa. gov/IAS/handbook_toc. html All materials by Austin Troy © 2003

Introduction to GIS LANDSAT 7 • Spatial resolution of bands Table 6. 1 Image Dimensions for a Landsat 7 0 R Product Band Number Resolution (meters) Samples (columns) Data Lines (rows) Bits per Sample 1 -5, 7 30 6, 6000 8 6 60 3, 300 3, 000 8 8 15 13, 200 12, 000 8 Source: http: //ltpwww. gsfc. nasa. gov/IAS/handbook_htmls/chapter 2. html All materials by Austin Troy © 2003

Introduction to GIS LANDSAT 7 • Spatial resolution of bands Band wavelength spectrums are slightly different from LANDSAT 5 All materials by Austin Troy © 2003

Introduction to GIS LANDSAT 7 • LANDSAT 7 has an excellent mission coverage archive Source: http: //ltpwww. gsfc. nasa. gov/IAS/handbook_htmls/chapter 6. html All materials by Austin Troy © 2003

Introduction to GIS LANDSAT Products • All data older than 2 years return to "public domain" and are distributed by the Earth Resource Observation System (EROS) Data Center of the US Geological Servey • Available at http: //edcwww. cr. usgs. gov/products/satellite/landsat 7. html • The LANDSAT Reference system catalogues the world into 57, 784 scenes, each 115 miles (183 kilometers) wide by 106 miles (170 kilometers) long. All materials by Austin Troy © 2003

Introduction to GIS Landsat Products: USGS • OR: no radiometric or geometric correction applied. Scan lines are reversed and nominally aligned. • 1 R: includes radiometric correction, but no geometric correction. Scan lines are reversed and nominally aligned. • 1 G: includes both radiometric and geometric correction. The scene will be rotated, aligned, and georeferenced to a user-defined map projection. All materials by Austin Troy © 2003

Introduction to GIS Landsat Products • Another source is Space Imaging Inc. • Note that they have resampled to 15 m All materials by Austin Troy © 2003

Introduction to GIS LANDSAT Imagery All materials by Austin Troy © 2003

Introduction to GIS LANDSAT Imagery Composite of shortwave infrared, Near-Infrared and Red. Shows manmade features as well as densely forested areas and agricultural lands All materials by Austin Troy © 2003

Introduction to GIS LANDSAT Imagery Same bands: shows wetlands, urban, open water, forest All materials by Austin Troy © 2003

Introduction to GIS LANDSAT Imagery Same bands: light yellow-green color represents northern hardwood forest. The dark green patches represent various conifer species All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • High resolution satellite developed by Space Imaging, launched 1999 • Has sun-synchronous orbit and crosses equator at 10: 30 AM • Ground track repeats every 11 days • Highly maneuverable: can point at a new target and stabilize itself in seconds, enabling it to follow meandering features • The entire spacecraft moves, not just the sensors All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • Can collect data at angles of up to 45° from the along track and across track axes • This allows for side by side and fore and aft stereoscopic imaging • At its nadir it has 11 km swath width • 11 km by 11 km image size, but user specified strips and mosaics can be ordered • Employs a linear array scanner All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • IKONOS collects panchromatic band (. 45 to. 90 mm) at 1 m resolution • Collects four multispectral bands at 4 m resolution • Bands include blue (. 45 to. 52 mm) , green (. 51 to. 60 mm) , red (. 63 to. 70 mm), near IR (. 76 to. 85 mm) • Radiometric resolution is 11 bits, or 2048 values All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • Here is 1 m IKONOS view of suburbs, near winter Olympics Source: spaceimaging. com All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • 1 m IKONOS view of Dubai All materials by Austin Troy © 2003 Source: spaceimaging. com

Introduction to GIS IKONOS data • 1 m IKONOS pan image of Rome Source: spaceimaging. com. All materials by Austin Troy © 2003

Introduction to GIS IKONOS data • 1 m image of “Survivor” camp in Africa Source: spaceimaging. com. All materials by Austin Troy © 2003

------Using GIS-- Introduction to GIS Part 2: Remote Sensing Image Processing and Interpretation All materials by Austin Troy © 2003

Introduction to GIS Image Pre-Processing • Once an image is acquired it is generally processed to eliminate errors • Two categories: • Geometric correction • Radiometric correction All materials by Austin Troy © 2003

Introduction to GIS Geometric Correction • Sources of distortion • variations in altitude • variations in velocity • earth curvature • relief displacement • atmospheric refraction • Skew distortion from earth’s eastward rotation All materials by Austin Troy © 2003

Introduction to GIS Geometric Correction • Raw digital images contain two types of geometric distortions: systematic and random • Systematic sources are understood and can be corrected by applying formulas • Random distortions, or ‘residual unknown systematic distortions’ are corrected using multiple regression of ground control points that are visible from the image All materials by Austin Troy © 2003

Introduction to GIS Radiometric Correction • Radiance measured at a given point is influenced by: • Changes in illumination • Atmospheric conditions (haze, clouds) • Angle of view • Instrument response characteristics • elevation of the sun (seasonal change in sun angle) • Earth-sun distance variation All materials by Austin Troy © 2003

Introduction to GIS Image enhancement • For improving image quality, particularly contrast • Includes a number of methods used for enhancing subtle radiometric differences so that the eye can easily perceive them • Two types: point and local operations • Point: modify brightness value of a given pixel independently • Local: modify pixel brightness based on neighborhood brightness values All materials by Austin Troy © 2003

Introduction to GIS Image enhancement • Three types of manipulation are: • Contrast enhancement: methods include gray level thresholding, level slicing and contrast stretching • Spatial feature manipulation: methods include spatial filtering, edge enhancement and Fourrier analysis • Multi-image manipulation: methods include multispectral band ratioing and differencing, principal components, canonical components, vegetative components, decorrelation stretching, others All materials by Austin Troy © 2003

Introduction to GIS Contrast enhancement (point operation) • Most images start with low contrast; these improve it • Level slicing reclasses DNs into fewer classes, so differences can be more easily seen; colors or grayscale values can be assigned. Like resampling down radiometric resolution. Often used where histogram shows bimodal distribution of reflectance values • Contrast Streching is the opposite, where a smaller number of values are stretched out over full DN range All materials by Austin Troy © 2003

Introduction to GIS Contrast enhancement • Here is what spectral histograms look like Note that DN is not zero for any of them All materials by Austin Troy © 2003 Source: http: //www. sci-ctr. edu. sg/ssc/publication/remotesense/process. htm

Introduction to GIS Contrast enhancement • The image on the left is hazy because of atmospheric scattering; the image is improved (right) through the use of Gray level thresholding. Note that there is more contrast and features can be better discerned All materials by Austin Troy © 2003 Source: http: //www. sci-ctr. edu. sg/ssc/publication/remotesense/process. htm

Introduction to GIS Spatial Feature Enhancement (local operation) • Spatial filtering/ Convolution: neighborhood operations (like we reviewed for raster analysis), that calculate a new value for the center pixel based on the values of its neighbors within a window (see “More Raster Analysis” lecture for more); includes low-pass (emphasizes regional spatial trends, demphasizes local variability ) and high-pass (emphasizes local spatial variability) filters All materials by Austin Troy © 2003

Introduction to GIS Spatial Feature Enhancement • Edge Enhancement: This is a convolution method that combines elements of both low and high-pass filtering in a way that accentuates linear and local contrast features without losing the regional patterns • First, a high-pass image is made with local detail • Next, all or some of the gray level of the original scene is added back • Finally, the composite image is contrast stretched All materials by Austin Troy © 2003

Introduction to GIS Image classification • This is the science of turning RS data into meaningful categories representing surface conditions or classes • Spectral pattern recognition procedures classifies a pixel based on its pattern of radiance measurements in each band: more common and easy to use • Spatial pattern recognition classifies a pixel based on its relationship to surrounding pixels: more complex and difficult to implement • Temporal pattern recognition: looks at changes in pixels over time to assist in feature recognition All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Two types of classification: • Supervised: the analyst designates on-screen “training areas” known land cover type from which an interpretation key is created, describing the spectral attributes of each cover class. Statistical techniques are then used to assign pixel data to a cover class, based on what class its spectral pattern resembles. • Unsupervised: automated algorithms produce spectral classes based on natural groupings of multi-band reflectance values (rather than through designation of training areas), and the analyst uses references data, such as field measurements, DOQs or GIS data layers to assign areas to the given classes All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Unsupervised: • Computer groups all pixels according to their spectral relationships and looks for natural spectral groupings of pixels, called spectral classes Spectral class 1 Spectral class 2 • Assumes that data in different cover class will not belong to same grouping • Once created, the analyst assesses their utility Source: F. F. Sabins, Jr. , 1987, Remote Sensing: Principles and Interpretation. All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Unsupervised: • After comparing the reclassified image (based on spectral classes) to ground reference data, the analyst can determine which land cover type the spectral class corresponds to • Has advantage over supervised classification: the “classifier” identifies the distinct spectral classes, many of which would not have been apparent in supervised classification and, if there were many classes, would have been difficult to train all of them. Not required to make assumptions of what all the cover classes are before classification. • Clustering algorithms include: K-means, texture analysis All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Unsupervised: • Here’s an example Source: http: //elwood. la. asu. edu/grsl/lter/fig 5. html All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Unsupervised: Another example Source: http: //mercator. upc. es/nicktutorial/Sect 1/nicktutor_1 -14. html All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Supervised: • Better for cases where validity of classification depends on a priori knowledge of the technician • Conventional cover classes are recognized in the scene from prior knowledge or other GIS/ imagery layers • Therefore selection of classes is pre-determined and supervised • Training sites are chosen for each of those classes • Each training site “class” results in a cloud of points in n dimensional “measurement space, ” representing variability of different pixels spectral signatures in that class All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Supervised: Here a bunch of pre-chosen training sites of known cover type All materials by Austin Troy © 2003 Source: http: //mercator. upc. es/nicktutorial/Sect 1/nicktutor_1 -15. html

Introduction to GIS Spectral Classification • Supervised: • The next step is for the computer to assign each pixel to the spectral class is appears to belong to, based on the DN’s of its constituent bands • There are numerous algorithms the computer uses, including: • Minimum distance to means classification (Chain Method) • Gaussian Maximum likelihood classification • Parallelpiped classification All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Supervised: • These algorithms look at “clouds” of pixels in spectral “measurement space” from training areas, and try to determine which “cloud” a given non-training pixel falls in. • The simplest method is “minimum distance” in which a theoretical center point of point cloud is plotted, based on mean values, and an unknown point is assigned to the nearest of these. That point is then assigned that cover class. • They get much more complex from there. All materials by Austin Troy © 2003

Introduction to GIS Spectral Classification • Supervised: • Examples of two classifiers All materials by Austin Troy © 2003 Source: http: //mercator. upc. es/nicktutorial/Sect 1/nicktutor_1 -16. html

Introduction to GIS Classifying Imagery • Spectral classification can be used for numerous purposes, like classifying geology, water temperature, soil moisture, other soil characteristics, water sediment load, water pollution levels, lake eutrophication, flood damage estimation, groundwater location, vegetative water stress, vegetative diseases and stresses, crop yields and health, biomass quantity, net primary productivity, forest vegetation species composition, forest fragmentation, forest age (in some cases), rangeland quality and type, urban mapping and vectorization of manmade structures • One of the most common applications of classification is land cover and land use mapping All materials by Austin Troy © 2003

Introduction to GIS Land Cover/ Land Use Mapping • Land cover refers to the feature present and land use refers to the human activity associated with a plot of land • The LU/LC classes to be derived will depend on the system being used. One of the most common is the USGS Anderson Classification System (Anderson et al. 1976). This classification scheme is hierarchical, with nine very general categories at Level I, and an increasing number of classes and detail and level increases. Paper available online at http: //landcover. usgs. gov/pdf/anderson. pdf • Anderson system intermixes land use and land cover metrics, by inferring land use from land cover. Unfortunately, land cover can only tell us a limited amount about land use—think of outdoor recreation as a land use. Need additional data for these classes. All materials by Austin Troy © 2003

Introduction to GIS Land Cover/ Land Use Mapping • Land use and land cover classification system for use with remote sensor data (Anderson et al. 1976) • Level II • 1 Urban or Built-up Land 11 Residential 12 Commercial and Services 13 Industrial 14 Transportation, Communications, and Utilities 15 Industrial and Commercial Complexes 16 Mixed Urban or Built-up Land 17 Other Urban or Built-up Land All materials by Austin Troy © 2003

Introduction to GIS • Level I Land Cover/ Land Use Mapping • 2 Agricultural Land Level II 21 Cropland Pasture 22 Orchards, Groves, Vineyards, Nurseries, and Ornamental Horticultural Areas 23 Confined Feeding Operations 24 Other Agricultural Land • 3 Rangeland 31 Herbaceous Rangeland 32 Shrub and Brush Rangeland 33 Mixed Rangeland • 4 Forest Land 41 Deciduous Forest Land 42 Evergreen Forest Land 43 Mixed Forest Land • 5 Water 51 Streams and Canals 52 Lakes 53 Reservoirs 54 Bays and Estuaries • 6 Wetland 61 Forested Wetland All materials by Austin Troy © 2003 62 Nonforested Wetland

Introduction to GIS Land Cover/ Land Use Mapping • Level II • 7 Barren Land 71 Dry Salt Flats. 72 Beaches 73 Sandy Areas other than Beaches 74 Bare Exposed Rock 75 Strip Mines Quarries, and Gravel Pits 76 Transitional Areas 77 Mixed Barren Land • 8 Tundra 81 Shrub and Brush Tundra 82 Herbaceous Tundra 83 Bare Ground Tundra 84 Wet Tundra 85 Mixed Tundra • 9 Perennial Snow or Ice 91 Perennial Snowfields 92 Glaciers All materials by Austin Troy © 2003

Introduction to GIS Land Cover/ Land Use Mapping • Level 3 and 4 categories deliver even more detail. • USGS only specifies classifications for 1 and 2. They suggest that higher level classification be designed by local planners who know the land uses, because of the narrowness of the categories • As an example for level 3, with “urban” (level 1) “residential” (level 2) category, includes single family home (111), multifamily home (112), group quarters (113), mobile home parks (115), etc. • LANDSAT data can be used to generate level 1 easily, level 2 with some finesse (15 to 20 m resolution recommended) • Levels 3 and 4, IKONOS data or aerial photographs are needed. Level 4 requires much supplemental information All materials by Austin Troy © 2003

Introduction to GIS Land Cover/ Land Use Mapping • Here is an example of LANDSAT data classified using the Anderson System All materials by Austin Troy © 2003

Introduction to GIS Accuracy Assessment • This is one of the most important parts of image classification. • Error rates can be very high in classification accuracies, especially with lower resolution data, and where pixels are mixed • This is often the most time consuming part of image classification • NLCD effort undertook effort to classify errors in each type of land cover, broken down by region of the US • User’s accuracy for type X: Percent of pixels classified as X that really are X. Producer’s accuracy: percent of pixels that were classified as other than X but really are X. All materials by Austin Troy © 2003

Introduction to GIS Object-Oriented Classification Traditional classifiers don’t work as well for new generation of high resolution data, like this 2 foot Emerge Color infrared airphoto. Why? Meaningless to classify each pixel All materials by Austin Troy © 2003

Introduction to GIS Object-Oriented Classification • This is one of the newer methods of image classification, designed for high-res data. Looks at the spatial grouping of pixels with similar reflectance characteristics and can create polygon objects representing homogeneous areas. • Allows for complex automated classification rules to be set based on both the spectral and spatial properties of the data • Using this information, objects can be classified based on tone, texture, shape and context • Allows for nested hierarchical classifications, from general to highly specific. All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps: • Segmentation of rasters into polygon objects • Objects are defined such that they minimize within-unit heterogeneity and maximize between unit heterogeneity, subject to some user defined parameters. • The user can control the scale parameter for acceptable level of heterogeneity. They can also control the degree to which segmentation is based on spectral or spatial characteristics, since heterogeneity is defined in terms of both. By repeating the segmentation with different scale parameters, the user can create a nested hierarchy of objects>>big objects containing smaller objects, containing smaller objects All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps: • Two levels of segmentation Source/More info: see Ecognition website: http: //www. definiens-imaging. com/index. htm All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps: • Following segmentation, each object is encoded with information about its tone, shape, area, context, neighborhors and spectral characteristics (e. g. mean, standard deviation, max, min or each band’s spectral reflectance) • This information can be used for feature extraction in which objects’ properties are analyzed to look for characteristics that help to discriminate one object type from another. That is, what object information helps discriminate one from another? All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps: • Then objects are classified by either defining training areas of known cover type (known as supervised fuzzy classification) or creating class descriptions organized through inheritance-based rules into a knowledge base (known as fuzzy knowledge base classification). All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps: • In the knowledge base approach, complex membership functions can be derived that describe characteristics that are typical or atypical for a certain class. The more a given object displays the characteristics, the more likely it is to be classified into the class to which those characteristics pertain. Characteristics can be based on spectral response summary statistics, shape characteristics, adjacency, connectivity, and overlay with certain thematic features. All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps • The classification can be hierarchical and nested, with finer classifications within coarser ones • Small classified objects can be aggregated up to large object classes and large objects can be split into smaller ones. Can then assign different segmentations to different class hierarchy level • Allows for high precision classifications within coarser, general classifications All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps • The classification can be hierarchical and nested, with All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps • Can use additional thematic layers to populate the knowledge base and create rules about what a certain class can be on top of, next to, or near. This can increase the accuracy of classifications, especially as you increase categorical precision and start getting into classifiying land uses in addition to land cover • Hence, when you do training areas, you not only get average spectral responses and shape metrics for a class, but also can get average values from underlying layers to help increase classification accuracy • Examples: farm fields as fn of slope, soils, etc; different suburban development types as function of distance to urban centers, income, crime, etc. All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Steps • Error evaluation: Can then assess the strength of each classification to see how well objects fall into them • However, if a pixel had a high fit score for two categories, it is “unstable. ” Here are the mean stability scores • Stability does down as number of categories goes up • This does not tell you classification, accuracy, which requires ground truth data All materials by Austin Troy © 2003

Introduction to GIS Object-oriented classification Software • e. Cognition: one of the top Object oriented classification software packages More info: see Ecognition website: http: //www. definiens-imaging. com/index. htm All materials by Austin Troy © 2003