Chapter 1 Introduction to Digital Image Processing of Remotely
Major Characteristics of Remote Sensing systems, Table 1-2; p. 13 Image Processing considerations, Fig. 1-3;.
Chapter 2 Remote Sensing Data Acquisition Alternatives
p. 26-37 Characteristics of Multispectral Remote Sensing systems, Table 2-2; 2-4; p. 40-42 Characteristics of
Thematic Mapper Spectral Bands, Table 2-6; Fig. 2-24, 2-25
p. 60-61 Digital Image Data Formats
Chapter 3 Image Processing System Considerations
Chapter 4 Initial Statistics Extraction
Chapter 5 Initial Display alternative and Scientific Visualization
Fig 5-8, 5-9; 5-11;
p. 101 RGB to IHS Transformation and Back Again, Fig 5-14,15;
Chapter 6 Image Preprocessing: radiometric and geometric correction
Fig. 6-1, 6-3; 6-6,7;
Chapter 7 Image Enhancement
p. 152 Band Ratioing
Chapter 8 Thematic Information Extraction: Image Classification
Chapter 9 Digital Change Detection
Chapter 10 Geographic Information Systems
p. 282 Data Structure; Vector Data Model; Raster Data Model; Quadtree Raster Data Model; DEM; TIN; (TOSCA).
Remote Sensing Applications in the Geosciences - Course Notes
The electromagnetic spectrum, 7,17
Photographic; Thermal IR; Rador K, X, and L bands
Elements of photo interpretation 2, 5, 9
Tone; colour or relative brightness
texture; frequency of tonal change produced by the imaging of aggregations of unit forms which are too small to be detected individually but give rise to repetitive structure in the image. Pattern and texture differ in terms of scale of observation.
pattern; spatial arrangement of objects
size and association
Data processing - enhancement; decorrelation; multivariate analysis 10
Visible spectrum 400-700 nonometers (.5 micrometers) 8, 17
Spot Green 10x10 metres: .5-.59 Red .61-.68 Near-IR .79-.89 24, 25
Landsat 60x80 metres: MSS 1 .5-.6 green; 2 .6-.7 red; 3 .7-.8 Near IR; 4 .8-1.1 Near IR26, 27
TM 30x30 metres: .45-.53 blue; .52-.57 green; .63-.69 red; .76-.9; Near IR; 1.55-1.75 SWIR; 2.08-2.35 SWIR; 10.8-12.9 TIR (Thermal Infrared, 120x120 metres)
DATA OR IMAGE ANALYSIS
Data Enhancement, facilitates visual interpretation 58
Data Classification, data reduction placing features with similar spectral properties into one of a set of defined categories. 58
DEMS (Digital Elevation Model)
Visual representations of the digital image are known as domains. 59
Image domain, visual image
Frequency domain, representation of the data in the form of histograms
Feature Space domain, the coordinate system is determined by the number of bands projected and their radiometric resolution; used to develope interpreation keys.
Image Correction 60
- Resampling techniques 64
Nearest neighbour, transfer of gray level of nearest pixel, simple and gray levels unaltered; image blocky, pixels offset by up to half a pixel.
Bi-linear Interpolation, transfer proximity weighted average of 4 nearest pixels, smoother and more accurate image, gray values are altered; some blurring.
Cubic convolution, transfer evaluated weight of 16 nearest pixels, very smooth image and most accurate; gray values are altered, slow.
Contrast stretch using LUT 68
Linear; histogram or frequency stretch (display range assigned is proportional to the frequency[no of pixels] in the histogram segment; infrequency stretch (inverse of frequency stretch); logarithmic stretch, mid-range brightness is enhanced and histogram skew corrected; custom breakpoint stretch, deductive approach. 70, 71
Gray Level Thresholding, to distinguish land and water. 72
In Photo-Paint, pixels lighter than a specified threshold value become white, darker pixels become solid.
Edge enhancements, enhance the visibility of existing edge features; uses a boxcar filter of 5x5 pixels; pixel brighness values are selectively altered to mark location of significant changes in brightness slope. Produces some degradation in general image contrast and quality; most valuable where information on structures are of critical importance (Adaptive Unsharp; unsharp - edge and a certain amoutn of smoothness).
Ratio Images use brightness ratios of different wavelength bands to enhance target/background contrast. Ratio values must be scaled to produce an acceptable qulity image for visual interpreation. 74
SUPERVISED CLASSIFICATION 75
Uses signals from specified training sites.
UNSUPERVISED CLASSIFICATION 76
Assigns cover types to classes based on spectral values alone; done on the basis of tone and other visual elements such as pattern and texture are not considered.
PRINCIPLE COMPONENT ENHANCEMENT 78
Individual bands are often highly correlated. Correlation can be checked by looking at scatter plots. Principal component enhancment can be used to identify new axes which maximize variance
DATA INTEGRATION 80
Integration of Remotely sensed data with geologic, geophysical
and geochemical data 82
Step 1 enter the digitized data into an Image Analysis System (e.g. Image Works)
Step 2 Registration and Geometric correction (Geocoding); data pre-processing
Step 3 Enhancements and integration to ANALYSIS, then EXTRACTION, then addition to a GIS obtained by digitizing maps; spatial analysis and modelling of geologic data.
SPATIAL ANALYSIS of combined data images.
The ultimate goal is to produce geocoded colour images in which the colours can be meaningfully interpreted and related to terrain/geologic features ... this implies that the reflection or emission characteristics of the input data are known and that the particular integration technique selected preserves the relationship between the input data characteristics and the resulting image colours - the colours are meaningful!! 84
SAR (Synthetic Aperture Sensor). 86
Raster integration techniques 89
Band combinations - one band per red green and blue guns of the CRT.
Colour Space Transforms - transformation of data into a differetn display space-colours can be quantifiably described by orthogonal axes, (intensity, saturation, hue) in display space (RGB triangle has red, green, and blue at the apices). 89
COLOUR COMPOSITE 94
RGB and IHS Colour spaces 100
IHS Black to white on vertical axis = Intensity; black to colour on horizontal axes = saturation; points of a horizontal polygon are red, yellow green, cyan, blue, magenta hues.
The intensity vector can represent a gray value for the radar channel
Magnetic data can be represented by the hue, and radar by the intensity.
Saturation value can be set to an arbitrary value to ephasize the feature represented by the hue.
It can be set to ensure a balance between hue and intensity; geology as hue and radar image as intensity.
SPECTRAL GEOLOGY 144
SPECTRAL REFLECTANCE 146
TM PROCESSING TECHNIQUES FOR DISCRIMINATING LITHOLOGY AND ALTERATION 158
Decorrelation stretch see page 10
PRINCIPAL COMPONENT ANALYSIS
ALTERED ROCK SPECTRA, GOLDFIELD, NEVADA 160
STRUCTURAL GEOLOGY 168
Idrisi Tutorial Exercises
1. The IDRISI for Windows environment
2. The display system
3. Map composition
4. Palettes, symbols and scaling
5. Database query
6. Map algebra
7. Database workshop
8. Distance and context operators
9. Automating analyses with macros
10. Cost distances and least cost pathways
11. Image exploration
12. Supervised classification
13. Principal components analysis
14. Unsupervised classification
15. Image georegistration using RESAMPLE
16. Digital cartographic databases
17. Changing reference systems with PROJECT.
PCI - Using PCI Software, V. 1, 2
1. User Interfaces
2. Video Display: concepts and use
3. Database Management
5. Importing data
6. Preprocessing and Geometric correction
7. Enhancements: contrast manipulation
8. Enhancements: spatial filtering
9. Enhancements: multi-image manipulations
10. Image Classification
11. Working with vectors
12. Working with Attribute data
13. Generating and Working with DEMs
14. Orthorectification and DEM extraction
15. Hyperspectral Analysis
16. SAR (Synthetic Aperture) analysis
17. Spatial Modelling: Raster GIS
18. Data Presentation
19. Exporting and Archiving Data
App. A. DCP: Display Control Program
App. B Glossary of Terms
ER Mapper 5.0 Tutorial
Table of contents
1. Introduction ot ERMAPPER -1
2. User interface basics - 11
3. Creating an algorithm -27
4. Working with data layers - 43
5. Viewing image data values - 59
6. Enhancing image contrast - 67
7. Using spacial filters - 89
8. Using formulas - 101
9. Geolinking images - 119
10. Writing images to disk - 137
11. Colourdraping images (Translucent layers) - 145
12. Mosaocking images - 159
13. Virtual datasets - 175
14. 3-D perspective viewing - 191
15. Thematic raster overlays - 201
16. Composing maps - 213
17. Unsupervised classification - 229
18. Supervised classificiation - 241
19. Raster to vector conversion - 259
A System setup - 269
B Reference texts - 271
Index - 273
Raster data includes satellite and aerial images, DTM’s and geophysical data
An ERM data set has two files: a binary BIL file and a header .ers file.
Vector data is stored as lines, points and polygons; as an ASCII data file and an .erv ASCII header file.
Display format is called colour mode, RGB or HSI
Also involves the display of statistical information about the image, e.g. histograms.
Removal of geometric errors, alignment with real world projections, and the geometric alignment of two or more images.
assembly of several adjacent images into a single image.
Image enhancement (processing)
Image merging, e.g. TM and SPOT; colour draping, e.g. vegetation over gravity as z; contrast enhancement; filtering; formula processing (algebraic manipulation); classification.
Dynamic Links overlays
Links to other file formats without need to convert files.
Annotation and map composition
Add vector data by drawing directly on screen.
Data export and hardcopy
Image Processing Tasks - p. 5
Each Image file(s) is contained in its own
directory, and the files referred to in the directories are algorithms
used to manipulate the image.
Mouse manipulation e.g. double-click to select, drag the title bar, resize window; GUI tools, e.g. zoom, select active window; and commands, e.g. FILE menu Open, are mostly standard windows-like operations. However note:
zoom uses a marquee
click the image with the left mouse button to pan such that the clicked point becomes the centre of the image;
shift-click to zoom out;
use View and Quick zoom to zoom to extents, and View Geoposition to use a set of zoom buttons;
Geographic Information Systems for Geoscientists - Bonham-Carter
p. 1-3. GIS is simply a computer system for managing spatial data. GIS have capabilities for data capture, input, manipulation, transformation, visualization, combination, query, analysis, modelling and output. The ultimate purpose is to provide support for making decisions based on spatial data.
Use a GUI or a command language.
Data is geocoded, which means it is geographically located.
Different geocoded data sets are spatially registered, that is they overlap correctly.
There may be a straight line constant ratio relationship between two elements, but the spatial distribution of the primary values may be nodal or random or linear contoured or patterned, etc. Or the ratio may be bimodal both graphically and spatially.
p. 3 Visualization reveals spatial patterns.
p. 4 Spatial query allows the asking of the questions :
What are characteristics of this location?
Whereabouts do these characteristics occur? (Where do gold and sulphur occur together?)
p. 5 Combination merges spatial datasets, e.g. a geological map and a satellite image.
p. 5 Analysis e.g. trend surface analysis.
p. 6 Prediction e.g. what are the parameters defining sites of gold mineralization.
p. 14-22 A Model GIS Study for Mineral Potential Mapping.
p. 32-39 Raster and Vector Spatial Data Models
p. 39-43 Attribute Data.
p. 43-49 The Relational model.
p. 52-68 Raster Structures.
p. 68-81 Vector Data Structures (TOSCA)
p. 87-90 Map Projections.
p. 95-101 Digitizing.
p. 103-108 Coordinate Conversion.
p. 120-126 Colour, Colour lookup tables.
p. 186-202 Map Reclassification.
p. 204-210 Operations on Spatial Neighbourhoods.
p. 267 271 Map Analysis, Types of Models.
p. 272 Boolean Logic Models, Landfill Site Selection, Mineral Potential Evaluation.