MULTI-TEMPORAL OBJECT-ORIENTED CLASSIFICATIONS AND URBAN ANALYSIS. CASY STUDY BELO HORIZONTE - MINAS GERAIS STATE Dr. Hermann Johann Heinrich Kux Msc. Eduardo Henrique Geraldi Araújo 1 st MULTIDISCIPLINARY WORKSHOP ON EXTRACTING AND CLASSIFYING URBAN OBJECTS FROM HIGH RESOLUTION SATELLITE IMAGES - 2007
27
Embed
MULTI-TEMPORAL OBJECT-ORIENTED CLASSIFICATIONS AND URBAN ...sites.poli.usp.br/ptr/labgeo/workshop/Araujo-Kux_INPE.pdf · multi-temporal object-oriented classifications and urban analysis.
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
MULTI-TEMPORAL OBJECT-ORIENTED
CLASSIFICATIONS AND URBAN
ANALYSIS. CASY STUDY BELO
HORIZONTE - MINAS GERAIS STATE
Dr. Hermann Johann Heinrich KuxMsc. Eduardo Henrique Geraldi Araújo
1st MULTIDISCIPLINARY WORKSHOP ON EXTRACTING
AND CLASSIFYING URBAN OBJECTS FROM HIGH
RESOLUTION SATELLITE IMAGES - 2007
Are our cities well known?
� Observation on the bad use of urban areas:
� Inadequate growth of urban areas with
great potential;
� Lack of information for adequate urban
planning activities.
Motivation for the study
Examples
� Evaluate the performance and characteristics of object-based image classifications as a contribution to urban planning and to sustainable development, by the presentation of a case study: Belo Horizonte, MG, Brazil.
Objective
� Characterization of area under study;
� Object Based Image Analysis, Data preparation and Classification results;
� Conclusions
Summary
� Characterization of area under study;
� Object Based Image Analysis. Data preparation and Classification results;
� Conclusions
Summary
Area under study
Image 2004
Image 2002
Buritis Belvedere
Study Area
Aspects of area under study
� Characterization of area under study;
� Object Based Image Analysis. Data preparation and Classification results;
� Conclusions
Summary
Methodological Procedures
Geological risks
Orthorectification
Analysis of:
- Requirements;
- Relevance;
- Membership
DEFINITION OF
AREA UNDER STUDY
Quickbird image 2002
Information acquired
Quickbird image 2004
Contour lineGEOMETRIC
CORRECTIOND-GPS (GCP)
DEM
Edited CadastreCLASSIFICATION
Data preparation
Definition of classes
Segmentation
Hierarchy
Membership rules
Evaluation
Ortho images
Ancillary data
Land cover Urban growthSPATIAL
INFERENCES
ANALYSIS OF
RESULTS
CONCLUSIONS Information generated
Field survey
Blocks
Streets
Outside
Ortho images and Cadastre
Data Preparation
High brightness. No discrimination of probable building materials. 10) White cover
Cover of large buildings. Presents many variations.9) Gray cover
Roofs, mainly of new asbestos and cement. Quantization level close to 2048.8) Flare
Strong response in the blue and sometimes in the green band. Well defined.7) Swimming pool
Low brightness. Close to high buildings and arboreal vegetation.6) Shadow
Used from urban cadastre to delimit streets, avenues and roads. Linear and
straight forms
5) Asphalt
Earth works, prepared for constructions. No standard. Irregular forms, variable
texture and random localization.
4) Bare soil
Simple square geometry, orange-like color with large tone variation. Smooth
texture. Easily identifiable visually.
3) Ceramics tile
Also strong response in the IR band. Uniform. Texture is smoother than
Arboreal vegetation.
2) Grass
Strong response in the IR band. Texture due to different heights of trees
(shadow). Easily distinguishable from other classes.