Impacts of spatial resolution on land cover classification
Impacts of spatial resolution on land cover classification. Chanida Suwanprasit and Naiyana Srichai Prince of Songkla University Phuket Campus. APAN 33 rd Meeting 13-17 February 2012. 2/20. Outline. Introduction Objective Methodology Results Conclusions. 3/20.
Impacts of spatial resolution on land cover classification
E N D
Presentation Transcript
Impacts of spatial resolution on land cover classification ChanidaSuwanprasit and NaiyanaSrichai Prince of Songkla University Phuket Campus APAN 33rdMeeting 13-17 February 2012
2/20 Outline • Introduction • Objective • Methodology • Results • Conclusions
3/20 Spatial Resolution is a measurement of the spatial detail in an image, which is a function of the design of the sensor and its operating altitude above the Earth’s surface (Smith, 2012). Classification Factors • Number of mixed Pixel • Number of ROIs • Scale or spatial resolution • Spectral resolution • Temporal resolution
5/20 Objective • To examine effects of pixel size on land use classification in Kathu district, Phuket, Thailand
7/20 Study area: Kathu, Phuket Kamala Kathu Patong
6/20 Data set specification
10/20 Band 1 (Blue) Band 2 (Green) Band 3 (Red) Landsat 5 Spectral Bands Band 4 (NIR) Band 7 (MIR) Band 5 (NIR)
11/20 Band 1 (Red) Band 2 (Green) THEOS Spectral Bands Band 3 (Blue) Band 4 (NIR)
9/20 True Color Landsat 5 THEOS
13/20 RGB (4,3,2) Landsat 5 THEOS
Data Set 12/20 THEOS Landsat 5 Process Overview Unsupervised K-Mean • Classes • Forest • Built-up • Road • Water • Agriculture • Grassland • Bare land Control points Training area Supervised SVMs Test area Land use Classification Map THEOS LandSat 5
14/20 Landsat 5 THEOS Unsupervised Classification:K-Mean (7 Classes)
16/20 Landsat THEOS Support Vector Machines : SVMs Forest Built - up Bare land Grassland Road Water
17/20 Class Confusion Matrix
18/20 Conclusion • THEOS gave a higher classification accuracy than Landsat 5 for identifying land use in this study. • More Spectral bands from Landsat 5 with 30m is not appropriated for selecting clearly ROIs than THEOS with 15m resolution. • The better resolution image greatly reduce the mixed-pixel problem, and there is the potential to extract much more detailed information on land-use/land cover structures.
19/20 References • Duveiller, G. and P. Defourny (2010). "A conceptual framework to define the spatial resolution requirements for agricultural monitoring using remote sensing." Remote Sensing of Environment114(11): 2637-2650. • Randall B. Smith (2012). "Introduction to Remote Sensing Environment (RSE)". Website: http://www.microimages.com.
20/20 Acknowledgement • Faculty of Technology and Environment Prince of SongklaUniversity, Phuket Campus • Geo-Informatics and Space Technology Development Agency (Public Organization) • UniNet