Raster Filters

1
Lecture 03-04 Neighborhood Operations
Topics
Raster Filters
References
Chapter 7 in Chrisman 2002, (pp. 169-194)
2
Lecture 03-04 Neighborhood Operations
Outlines
1. What is an image An array of values
raster data layer (The Raster Values
Figure) Pixel and spatial resolution
Basic raster format (The Raster Format
Figure) Raster layer, image, and
picture Turn raster layer into image
(picture) (User 3dMapper (www.terrainAnalytics
.com) to demo it)
3
2. The purpose of raster filtering To
emphasize certain kind of info 3. The process of
raster filtering Convolution of a filter
with the original image (The Convolution
Figure) Basic issues 1)
neighborhood size (The Neighborhood Figure)
2) the filter (weight template, weight kernel)
3) methods used to compute the output
value (The Convolution Figure)
4) the edge problem (The Edge
Problem Figure)
4
4. Filter Design 1) Types of filters
High pass filters (edge detectors) Low
pass filters (average filters) 2) Design of
filters (specifying the weight kernel)
The morphological approach
(1) the size (horizontal, plain-view)
(2) shape a) plain-view
shape b) cross section shape
(3) unbiased
Examples (Valley and Ridge
Filter Figures)
5
5. Detecting features 1) The basic steps
a) determine the components b)
design the filters for each components
c) perform the computation for each component
d) merge the components e) make a
decision as to what is a valley 2)
Examples a) Detecting valleys b)
Detecting ridges
6
6. Discussion Filters are feature specific
7. For your practice a) design a set of
filters for valley of 3 pixel wide b)
design a filter for detecting a flat mountain
top with a size of 3 pixel on the
each side c) do your homework using the
following image and filters
(The Homework Figure)