Topic 11: Raster Analysis Flashcards
Explain image enhancement
- Done for aesthetic reasons
- 3 different kinds: grey-level thresholding, density slicing, contrast stretching
- grey-level thresholding: used to identify part of an image or data set
- density-slicing: identify multiple thresholds
- contrast stretching: manipulating the histogram
Explain the difference between point, neighborhood, and zonal operations with raster data
point: calculate new values for a raster grid cell by cell, similar to query
neighborhood: take a square, overlay it over the raster, the outside cells are averaged to calculate the middle square, done all over the raster
zonal: “window size” is variable and defined by zonal layer eg., calculation of mean values within each zone
Explain the use of neighborhood (convolution) operators for analyzing raster data
Low pass filters: filter out noise (shows general trends)
High pass filters: takes out general trends and shows noise
Used in Remote sensing texture analysis, DEM processing, Spatial filtering
Qualities of Rasters?
Raster have some qualities that impact our abilities to process them in a GIS environment
- Entities (pixels) are constant between layers
- Typically, there is only one attribute (the grid code) per entity in any given layer
- Entities and attributes exist regularly and continuously
Are raster analyses faster than vector analysis?
- Raster analyses tend to be FASTER than vector analysis, and the basic functions SIMPLER
- However, strung together into algorithms, raster processing can be very complex, and many types of transformations can ONLY occur in a raster environment
Raster Data Operations
- Many vector transformation operations do not apply in the raster environment
- Global raster operations - done on entire image - enhancements
What are the 3 basic types of raster operations?
- “point” or local operations
- Neighborhood or convolution operations
- Zonal operations
Purpose of image enhancement?
Improve the visual appearance and interpretability of an image
- contrast manipulation (Grey level thresholding, density slicing, contrast stretching)
Purpose of image thresholding?
Used to segment an input image into two classes
- one class for values below a specified DN
- One class for values above a specified DN
Often used to prepare a binary image to separate spectrally-distinct features for further analysis
Will use it a lot to identify portion of image or data set
What is density (level) slicing?
- DNs along the x-axis of the histogram are segmented into analyst-defined intervals (slice)
- Similar to thresholding, except it involves numerous classes
- Simplistic approach for image classification
What is contrast enhancement?
- Original DN values rarely extended over the entire output range of a display device
- Contrast enhancement techniques enhance or ‘stretch’ original data to accentuate contrast/ interpretability of image
- Depending on technique used, data integrity may be affected
- Manipulation to histogram, trying to be more analytical to get people to see it, more of visual thing
- image will have low contrast
What is contrast stretching?
- Can grab different parts of histogram and pull/stretch them
the shape is the same, however extends over entire range
we don’t use all digital numbers, however there is a greater range of them - vibrant colours
Contrast enhancement techniques?
- Most of the time we are looking for aesthetics
- Complete trial and error and what looks the best
Histogram and Lookup tables
Linear stretch
- any value that is the minimum is going to be output to y axis
Expotential stretch
- Enhancing parts of histogram with the most values
Logarithmic
- Enhancing parts of histogram with the least values
What are point operations?
- Operations that calculate new values for a raster grid on a cell by cell basis
simplest form: reclassification of existing raster values
- Actually a very powerful technique that operates similar to a query (vector) and can be combined with other point operations to accomplish many things
Point operations: Raster calculator
- Map algerbra involves mathematical or boolean operations to be calculated amongst various raster layers on a pixel by pixel basis
What is a convolution or neighborhood operation?
- Convolution involves the passing of a moving window over an image and the creation of a new layer
- Each pixel in the new layer is a function of the original pixel values underlying the convolution window as specified by the user
- Also known as ‘moving window’, ‘local’ or ‘filtering’ operations
‘High-Pass’ and ‘Low Pass’ Filters
Low-pass filters remove high frequency information (noise) and show general trends in the data (shows general trends)
High-pass filters remove the low frequency information and show the high frequency (shows noise)
Convolution Applications
Convolution operations are widely applied in a variety of raster applications
- Remote sensing texture analysis
- DEM processing
- Spatial filtering - enhance
What are zonal operations?
Similar to neighborhood operations, but the “window size” is variable, and defined by a zonal layer
operations available in a zonal context are similar to those used in neighborhood operations
Slope magnitude (Eyton 1991)
Simplest application of calculating slope
Azimuth
Tells which way the slope is pointing towards
Relative Radiance
- Trying to simulate a light angle
- Based on slope and azimuth, how much light would a pixel reflect? (Assuing it is a diffuse reflector)
Unidirectional vs multidirectional hillshade
Important for visualization
Multi: don’t have strong shadows
Uni: can overwhelm in high relief, can give alignment that seem strong that are not really there
