16-09-2016, 04:03 PM
1454996790-Presentation1Copy.pptx (Size: 3.21 MB / Downloads: 7)
Introduction
Interpreting images is very complex and challenging task.
The heart of the problem is that for any image there may be many probable interpretations and there is no way to determine the correct interpretations with absolute certainty.
Objectives
Hierarchical Object Based Image Analysis.
To investigate the synergistic use of hyperspectral and LiDAR data for optimum result.
Automatic extraction of urban buildings and tree crown.
Object Based Image Analysis(OBIA)
The human brain is an efficient learning, storage, and quick retrieval mechanism that can store various patterns and can instantaneously retrieve the patterns from our memory banks.
Thus an experienced image analysts resolve these ambiguities with a high degree of accuracy.
eCognition attempts to computerize this ability of human being into image analysis and automatic feature extraction.
Object based image analysis is a sub discipline of GIScience devoted to partitioning remote sensing imagery into meaningful image-objects
With object-oriented methods, not only the spectral information but also the shape, contextual and semantic information can be used to extract objects.
The object-oriented building extraction typically includes several steps: data pre-processing, multi-scale image segmentation, the definition of features used to extract buildings and tree crown, building and tree crown extraction, post-processing and accuracy evaluation
A new approach is offered by the multiresolution segmentation in the eCognition software for object oriented image classification . The system allows intelligent classification of aerial and satellite imagery in different scales simultaneously by different object layers and it has already been proven to be a satisfying alternative solution.
Multiple sources information integration, such as with LIDAR data or map data is one of its uniqueness.
The combination of high resolution images and high resolution elevation information in an object-oriented method promises good results for the automatic updating of GIS data sets.
The final result of object-oriented classification can be the form of vector polygons or in a raster form.
In this regard the product is more suitable for the GIS expectations and it overcomes the traditional standard canons of the thematic cartography.
Image Segmentation
Image segmentation is the first step and also one of the most critical tasks of image analysis
For segmentation, the basic algorithms like chessboard, contrast split , or multi-resolution segmentations usually are used. In this project the emphasize is given to multi-resolution segmentation.
It is a bottom-up segmentation algorithm based on a pair wise region merging technique.
It minimizes the average heterogeneity and maximizes their respective homogeneity.