cheriemuleh-131114155758-phpapp02.pdf

Embed Size (px)

Citation preview

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    1/18

    Usando ENVI para

    extraer elementos

    importantes desdeimágenes satelitales y

    datos LiDAR

    Cherie [email protected] 

    The information contained in this document pertains to software products and

    services that are subject to the controls of the Export Administration Regulations

    (EAR). The recipient is responsible for ensuring compliance to all applicable U.S.

    Export Control laws and regulations.

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    2/18

    Agenda

    >  Consideration of Data Availability and Usage

    >  Feature Extraction Methods

    >  Applying Methods to Extract Building Features>  Future Prospects for Building Feature Extraction

    Extracting Building Features from LiDAR + Optical Data

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    3/18

    Data Types

    > Color/IR Orthophotos

    > Multi/Hyperspectral

    > LiDAR

    > SAR

    Platforms

    > Aerial

    > Spaceborne

    > Terrestrial

    Prospects for future data 

    > Commercial UAVs

    An Abundance of Geospatial Data from which to

    Extract Features and Information

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    4/18

    Valuing Remotely Sensed Data as a Source for Features

    Imagery is not just a base map, but a source of rich informationthat geospatial analysts can use to solve complex problems.

    > Provide data availability over broadand inaccessible areas

    > Improve timeliness of data acquisition

    > Potentially greater accuracy

    > Automated feature extraction for

    reduction in manual digitization

    > Advanced geospatial analysis usingspectral image properties

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    5/18

    Extracting Information from Remotely Sensed Data

    Features of Interest

    > Vehicles

    > TransportationNetworks

    > Structures

    > Natural Features

    > Human Activity

    Limitations or

    Opportunities,

    Given the Data Type

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    6/18

    Needs for Feature Extraction

    >  Increased availability of high-

    resolution images

    >  Manual digitization

    >  Semi-automated solution ishighly desired

    Applications

    >  Defense and Security

    >  Transportation

    >  Urban planning and mapping

    Extracting Information from Remotely Sensed Data

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    7/18

    What is an object?

    • An object is a region of interest withspatial, spectral (brightness and color),and/or texture characteristics that definethe region

    • Pixels are grouped into objects, insteadof single pixel analysis

    • May provide increased accuracy anddetail for classification purposes

    Object-Based Image Analysis

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    8/18

    Building Extraction Methods using Geospatial Data

    Pixel by Pixel

    Group materials based on their

    reflectance response per pixel

    0.5

    0.0

    1.0

    0.5

    0.0

    1.0

    1 2 3 4 5 6

    0.5

    0.0

    1.0

            R

          e         f        l      e       c        t       a       n      c       e 

    Band

    0.5

    0.0

    1.0

    0.5

    0.0

    1.0

    1 2 3 4 5 6

    0.5

    0.0

    1.0

            R

          e         f        l      e       c        t       a       n      c       e 

    Band

    Soil

    Veg

    Water

    1

    65

    43

    2

    ImagePixels

    > (+) Good for large area-based FX with low-med resolution data

    > (-) Poor edge detection without good spectral/spatial resolution;challenging for building extraction

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    9/18

    Object-based Image Analysis

    Image

    Pixels

    Segmented

    Objects

    Complex

    Building

    Features

    Merged

    Segmented

    Objects

    > Computer vision technique involving image segmentation> Objects are classified into feature classes based contextual

    attributes: spatial, textural and spectral> Yields accurate building extraction; results and can be model-based

    Building Extraction Methods using Geospatial Data

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    10/18

    For Planning and Risk Identification> Land use planning> Zoning, taxation> Structure inventory> Material Identification

    Building Feature Extraction: An Important Aspect

    for Understanding an Operational Landscape

    For Post-event Response

    > Disaster assessment> Response planning> Reconstruction

    monitoring

    Buildings are key foundational

    data layers for GIS and critical

    to decision analytics 

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    11/18

    Extracting Features from LiDAR Point Clouds

    Features extracted from Point Clouds

    > Requires thicker point clouds> Based on 3D morphological filters> Proprietary or custom algorithms

    DSM DEM Height Model

    Features interpreted from

    derivative raster products> Multi-step process> Feature delineations from

    interpolated height values> Use results with object-based FX

    Feature identification: 3D point cloud visualization

    > Manual process, but familiar and expedient

    Building Extraction Methods using Geospatial Data

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    12/18

    Objective:> Efficiently extract building footprints> Use imagery to glean information about the structures that

    will provide situational awareness

    Applying Methods to Extract Building Features

    Combining Optical and LiDAR Data for Decision Support 

    Process:

    > 3D Feature Extraction from hi-res LiDAR tocapture building footprints

    > Conduct image processing routines usingbuildings as regions of interest

    Combine the best of

    what LiDAR and image

    processing have to offer 

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    13/18

    Applying Methods to Extract Building Features: LiDAR

    Use Advanced 3D Algorithms to Process LiDAR Data 

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    14/18

    Applying Methods to Extract Building Features: LiDAR

    3D LiDAR Extraction Vector and Raster Products 

    Classified Point Cloud

    Trees

    > Location, Elevation, Height, Radius

    Buildings

    > Location, Perimeter Vectors, Roof FaceVectors

    Power Lines

    > Power Line Vectors, Power Pole List, PowerLine Attachment Points

    Terrain

    > Digital Surface Model (Grid and TIN),Digital Elevation Model, Ground contours

    Valuable GIS

    Data Layers 

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    15/18

    Applying Methods to Extract Building Features: LiDAR

    Leverage Building Footprints and Elevation Products 

    Determine Height Model

    > Raster data for additionalprocessing/awareness ofobjects in the area

    DSM DEM Height ModelBuilding Vectors

    > Immediate asset inventory> AOIs for additional processing

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    16/18

    Applying Methods to Extract Building Features: Optical

    Image Analysis Methods Using LiDAR-derived Products 

    Topographic Modeling

    > Use raster height modeldata to determine roofslope & aspect on buildings

    Spectral Analysis

    > Apply object-based FX tomulti/hyperspectralimagery, using building

    footprint ROIs> Capture additional spectral,textural, spatial attributesfor additional analysisopportunities

    Height Model ROI Roof Angle and Slope

    ROISpectral Image Roof Composition

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    17/18

    October 23, 2013 17

    Future Perspective: Building Feature Extraction

    Better Data, Better Tools, Better Analysis Results… 

    Improved Point Cloud FX> Denser data> MSI/HSI Spectral attribution> Improved algorithms

    Improved Object-Based FX> Better quality imagery> Better OBIA models

    3D Visualizations & Modeling

    > Photorealism & accuracy> New 3D analysis methods

    Convergence of tools and

    methods will improve building

    FX, regardless of data type 

  • 8/21/2019 cheriemuleh-131114155758-phpapp02.pdf

    18/18

    Thank You

    © 2013 Exelis Visual Information Solutions, Inc.