Object Based Image Analysis Tools for Opticks Mohit Kumar, KS Rajan, Dustan Adkins http://osgeo.in/foss4g2012 1 OSGEO-India: FOSS4G 2012- First National Conference "Open Source Geospatial Resources to Spearhead Development and Growth” 25-27 th October 2012, @ IIIT Hyderabad
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Object Based Image Analysis Tools for Opticks
Mohit Kumar, KS Rajan, Dustan Adkins
http://osgeo.in/foss4g2012 1
OSGEO-India: FOSS4G 2012- First National Conference "Open Source Geospatial Resources to Spearhead Development and Growth” 25-27th October 2012, @ IIIT Hyderabad
Why object-based?
http://osgeo.in/foss4g2012 2
• Opticks is an open source, remote sensing application that supports imagery, video (motion imagery), Synthetic Aperture Radar (SAR), multi-spectral, hyper-spectral, and other types of remote sensing data.
• Opticks can also be used as a remote sensing software development framework. Developers can extend Opticks functionality using its plug-in architecture and public application programming interface
• http://opticks.org
• Object based approach is better than conventional per-pixel analysis as it deals with considerably reduced number of units. This approach is not that much sensitive to noise and hence is spatially consistent.
OPTICKS ?
Workflow
http://osgeo.in/foss4g2012 3
http://osgeo.in/foss4g2012 4
Input image(RGB)
Input image (CIELAB colour
space)
5 Dimensional feature space
Modes(local maximas) Clustering
Objects formed (backtracking the
modes)
Points in feature space collapsing
to one mode form an object.
Pruning (spatial Bandwidth)
Pruning (Spectral Bandwidth)
Pruning ( Minimum region area)
Image Segmentation (Meanshift)
Object attribution• Calculating textural, geometric and spectral features for the objects
made in the Segmentation step in a feature vector. • Area, Perimeter, Roundness, Compactness, Centroid, Contrast,
Coarseness, Direction, Roughness, Mean red, Mean green, Mean blue, std. deviation Red, std. deviation Green, std. deviation Blue.
http://osgeo.in/foss4g2012 5
Segmented Image For every object in the image
Initialize a vector having all 16 features
Calculate value for every feature and save in the vector.
Classification• Mahalanobis Distance• Di,j
2 = (x-µj)`S-1(x-µj)• The class which has the least Mahalanobis
distance to the object i is the class of that object.
http://osgeo.in/foss4g2012 6
Vectorization• Creates vector polygons for all connected regions of pixels in the
object image sharing a common pixel value.• Polygon features are created on the output layer, with polygon
geometries representing the polygons.• The class which has the least Mahalanobis distance to the object i is
the class of that object.
http://osgeo.in/foss4g2012 7
The Input Orbview3 (4m) data of a part of delhi (500X500) The output of the objects with area less than 100.
Output of object having area 100-200 and classified as building. The shapefile(.shp) displaying the objects