Scribd Logo
Upload

Welcome to Scribd!

  • Extraction of Shrimp Ponds Using Object Oriented Classification Vis-A-Vis Pixel Based Classfication

    Uploaded by

    Sayedur R Chowdhury
    40 views1 page
    This document discusses using object-oriented classification to identify shrimp farms using satellite imagery. Traditional pixel-based classification approaches have difficulty distinguishing between features with similar spectral responses, such as active shrimp farms versus shallow rivers or dry ponds versus bare land. Object-oriented classification using eCognition software segments the image into objects based on spectral values, shape, and texture. This allows integration of different object features and extraction of real-world objects like shrimp farms and ponds, regardless of their state in the farming cycle. The study demonstrates improvements in identifying and extracting shrimp farms and ponds using object-based classification of features of interest.

    Original Description:

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    This document discusses using object-oriented classification to identify shrimp farms using satellite imagery. Traditional pixel-based classification approaches have difficulty distinguishing between features with similar spectral responses, such as active shrimp farms versus shallow rivers or dry ponds versus bare land. Object-oriented classification using eCognition software segments the image into objects based on spectral values, shape, and texture. This allows integration of different object features and extraction of real-world objects like shrimp farms and ponds, regardless of their state in the farming cycle. The study demonstrates improvements in identifying and extracting shrimp farms and ponds using object-based classification of features of interest.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    40 views1 page

    Extraction of Shrimp Ponds Using Object Oriented Classification Vis-A-Vis Pixel Based Classfication

    Uploaded by

    Sayedur R Chowdhury
    This document discusses using object-oriented classification to identify shrimp farms using satellite imagery. Traditional pixel-based classification approaches have difficulty distinguishing between features with similar spectral responses, such as active shrimp farms versus shallow rivers or dry ponds versus bare land. Object-oriented classification using eCognition software segments the image into objects based on spectral values, shape, and texture. This allows integration of different object features and extraction of real-world objects like shrimp farms and ponds, regardless of their state in the farming cycle. The study demonstrates improvements in identifying and extracting shrimp farms and ponds using object-based classification of features of interest.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    of 1

    302 25

    th
    ACRS 2004 Chiang Mai, Thailand

    B-3.6 Data Processing
















    )<86%'8-323*7,6-14432(797-2+3&.)'836-)28)('0%77-*-'%8-32
    :-7:-74-<)0&%7)('0%77-*-'%8-32
    Ravi Chauhan

    , Nitin K.Tripathi

    Sayedur R.Chowdhury

    Asian Institute oI Technology, P.O. Box 4


    Klong Luang Pathumthani 12120,
    THAILAND
    Tel: 66-2-524-6392, Eax: 66-2-524-5597 Email: nitinktait.ac.th
    KEY WORDS: Object oriented classiIication, eCognition, IRS LISS, shrimp Iarm
    ABSTRACT: Rapid expansion oI coastal
    aquaculture/shrimp Iarming in several
    countries has necessitated the inventory and
    monitoring oI shrimp Iarms. These are
    essential tools Ior decision making on
    aquaculture development, including
    regulatory laws, environmental protection
    and revenue collection. In the context oI
    aquaculture development policies oI
    respective governments, much attention are
    Iocused on the identiIication and
    classiIication oI shrimp Iarms, oIten located
    in remote areas. This study examines
    classiIication oI shrimp Iarms using pan
    sharpened IRS LISS-III image oI the
    Kandleru creek area oI Andhra Pradesh,
    India. IRS satellite image is subjected to
    pixel based vis-a-vis object oriented
    classiIication Ior improving the
    identiIication oI shrimp Iarms in the area.
    Most traditional pixel-based classiIication
    approaches are based exclusively on the
    digital numbers oI the pixels taking only the
    spectral inIormation in the scene into
    account. It does not consider the visible
    characteristics like pattern, texture, shape oI
    a ground Ieature. Eeatures having similar
    spectral response become diIIicult to
    distinguish Irom each other and are oIten
    classiIied into the same Ieature type. Active
    shrimp Iarms mostly having water depth oI
    1-2 meters hardly are diIIerentiated Irom
    shallow rivers. Likewise, dry ponds under
    preparation, between crop cycles, or
    abandoned are not diIIerentiable Irom bare
    land or sandy patches depending on the
    spectral response. Pan sharpened IRS LISS
    image resolution supports visual
    identiIication oI the dykes and canals in a
    shrimp Iarm otherwise neglected by the
    traditional pixel based classiIication. Isodata
    unsupervised and supervised maximum
    likelihood classiIications were perIormed on
    the scene to extract the shrimp ponds both
    dry and Iilled. The above problem is Iound
    to hinder successIul extraction oI the ponds.
    An object-oriented classiIication is
    suggested in order to overcome this
    limitation. E-Cognition soItware allows the
    polygon based classiIication process. It is
    based on Iuzzy logic, allows the integration
    oI a broad spectrum oI diIIerent object
    Ieatures, such as spectral values, shape and
    texture. In contrast to classic image
    processing methods, the basic processing
    units oI object-oriented image analysis are
    image objects or segments and not single
    pixels. One motivation Ior the object-
    oriented approach is the Iact that, in many
    cases, the expected result oI most image
    analysis tasks is the extraction oI real world
    objects, proper in shape and proper in
    classiIication. The present study
    demonstrates improvements in the
    identiIication and extraction oI shrimp Iarms
    and ponds irrespective oI their states in the
    Iarming cycle using the object based
    classiIication oI the Ieatures oI interest only.

    You might also like