Presentation is loading. Please wait.

Presentation is loading. Please wait.

A Knowledge-based Approach for Reducing Cloud and Shadow Mingjun Song and Daniel L. Civco Laboratory for Earth Resources Information Systems (LERIS) Department.

Similar presentations


Presentation on theme: "A Knowledge-based Approach for Reducing Cloud and Shadow Mingjun Song and Daniel L. Civco Laboratory for Earth Resources Information Systems (LERIS) Department."— Presentation transcript:

1 A Knowledge-based Approach for Reducing Cloud and Shadow Mingjun Song and Daniel L. Civco Laboratory for Earth Resources Information Systems (LERIS) Department of Natural Resources Management & Engineering The University of Connecticut U-4087, 1376 Storrs Road Storrs, CT 06269-4087

2 ProblemProblem Completely cloud-free remotely sensed images are not always available, especially in tropical, neo-tropical, or humid climates, posing complications and perhaps serious constraints to image analysis

3 ObjectiveObjective Develop a knowledge- based method to produce a cloud and cloud shadow-free multitemporal image composite of neo- contemporary images

4 MethodologyMethodology Topographical normalization Multi-date Brightness correction Main image versus secondary image To detect those areas which are covered with cloud and shadow in the main image, and not having cloud or shadow in the secondary image Replace

5 First Study Area Eastern Madagascar Eastern Madagascar December 15, 2000 April 22, 2001

6 Topographical Normalization

7 Normalization Result

8 Multi-date Brightness Correction

9 Criteria Band 1: Cloud Band 4: Shadow Band Difference Rationale: If the difference is less than a threshold, it should be the same object in two dates images

10 Criteria: Shape Information Shadow and stream. eCognition, length/width Length/width: Bounding Box: Covariance Matrix:

11 Knowledge Base ParameterCloudShadow Output Value21 Band 1 (Main Image)> 41>= 0 Band 4 (Main Image)>= 0< 35 Band 1 (Secondary Image)< 33 Band 4 (Secondary Image)>= 0> 27 Band 1 Difference> 10>= 0 Band 4 Difference>= 0> 10 Length-to-Width Ratio>= 0< 9

12 Imagine Spatial Model

13 Cloud and Shadow Detection Expert 01_band1 00_band1 01_band4 00_band4 Band1_dif Band4_dif Obj_lw

14 Composite Replace

15 Second Study Area ETM 23 April 2001 ETM 23 April 2001 ETM 26 March 2000 ETM 26 March 2000 Central-Eastern Connecticut Central-Eastern Connecticut Thames River Thames River UConn

16 Multi-date Brightness Correction Sample_01Sample_00Sample_00_corr00_corr

17 CriteriaCriteria Band 1: Cloud Band 4: Shadow Band Difference

18 Contextual Information Difficulty: Cloud edge with urban area Shadow with water area Rationale: Cloud edge and shadow area should be accompanied by clear cloud area Buffer the clear cloud area

19 Knowledge Base ParameterCloudShadow Output Value21 Band 1 (Main Image)> 95>= 0 Band 4 (Main Image)>=0< 46 Band 1 (Secondary Image)< 221 Band 4 (Secondary Image)>= 0 Band 1 Difference> 10>= 0 Band 4 Difference>= 0> 10 Contexture in Clear Cloud==1

20 Cloud and Shadow Detection 01_band1 00_band1 Band1_dif Context Band4_dif 00_band4 01_band4 Expert

21 Composite 2001 ETM 2000 ETM Cloud & Shadow Area Cloud & Shadow Area

22 Conclusion Procedure of mosaic Knowledge-based Spectral, shape, contextural Easy and efficient Flexible Additional image needed for overlap areas

23 AcknowledgementAcknowledgement National Aeronautics and Space Administration Grant NAG13-99001/NRA-98-OES-08 RESAC- NAUTILUS, Better Land Use Planning for the Urbanizing Northeast: Creating a Network of Value- Added Geospatial Information, Tools, and Education for Land Use Decision Makers. Northeast Applications of Useable Technology In Land planning for Urban Sprawl

24 This presentation is available at resac.uconn.edu

25 A Knowledge-based Approach for Reducing Cloud and Shadow Mingjun Song and Daniel L. Civco Laboratory for Earth Resources Information Systems (LERIS) Department of Natural Resources Management & Engineering The University of Connecticut U-4087, 1376 Storrs Road Storrs, CT 06269-4087


Download ppt "A Knowledge-based Approach for Reducing Cloud and Shadow Mingjun Song and Daniel L. Civco Laboratory for Earth Resources Information Systems (LERIS) Department."

Similar presentations


Ads by Google