Object Based Classification using Kolmogorov Smirnov Distribution

Authors: LASYA VENIGALLA*, University of Texas, Dallas, Fang Qiu, Professor and Department Head
Topics: Remote Sensing, Land Use
Keywords: Object-based classification, Image Segmentation, Python, Remote Sensing
Session Type: Paper
Day: 4/6/2019
Start / End Time: 1:10 PM / 2:50 PM
Room: Buchanan, Marriott, Mezzanine Level
Presentation File: No File Uploaded


Object-Based Classification using Kolmogorov Smirnov Distribution Abstract: One important technique that defines remote sensing is the classification of an image. Classification of an image can be performed at two levels: Pixel-level and Object level. This study involves the classification of high-resolution WorldView-2 satellite data. This DigitalGlobe owned American satellite comes in eight different spectral bands with a spatial resolution of 1.84 meters. The study area chosen for the classification is located in Dallas, Texas. The analysis is carried out in two main steps. First, image segmentation is performed on the image to form objects. For this, the Mean Shift Image Segmentation function in the arcpy module is used. Second, classification is performed using Kolmogorov Smirnov Distribution. A cumulative histogram is generated for the training and the testing segments. KS distance is calculated between the trained samples and the unknown segments. The calculated distance between the trained and the unknown segment is used to assign the unknown segment to a defined class. The entire analysis is formulated in python using the arcpy and scipy modules. The main objective of the study is to automate the entire procedure so that it could be used as a plug-in for ArcMap or ArcGIS Pro. Keywords: Object-based classification, Image Segmentation, Python, Remote Sensing

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login