change dec ti on

9
This lecture reviews basic considerations when This lecture reviews basic considerations when change information is extracted from digital change information is extracted from digital remotely sensed data and some of the remotely sensed data and some of the  procedures to the study of change detection.  procedures to the study of change detection. Digital Change Detection Digital Change Detection Why do we study change detection? Why do we study change detection? It is believed that land It is believed that land - - use/land use/land- -cover change is a major cover change is a major component of global change with an impact perhaps component of global change with an impact perhaps greater than that of climate change. greater than that of climate change. It is important that such changes be inventoried It is important that such changes be inventoried accurately so that the physical and human processes at accurately so that the physical and human processes at work can be more fully understood. work can be more fully understood. 1. Cla ssi ficati on sche me 2. Geogra phi c study ar ea 3. Ti me per iod 4. Per pi xel or object oriented approach 5. Temporal and s patia l res olut ion 6. Phenol ogical cycl es 1. Classifi cation scheme 2. Geographi c st udy a rea 3. Ti me per iod 4. Per pi xel or object oriented approac h 5. Temporal and spati al re solution 6. Phenol ogical cycl es Considerations for change detection Considerations for Considerations for change detection change detection Classifying Features Classification s ystems can be broad such as wa ter/land, or detailed such as single f amily/multifamily dwelling. The scale of the image will of ten dictate how finely you can classify features. The USGS ha s devel oped a sta ndard land use/land cover classificati on system based in the works of Anderson 1976. Other c lass ifica tion s ystems have been devel op by:  GLUT (Georgi a landuse tre nds )  NARSAL (  Natural Resources Spatial Analysis Laboratory (NARSAL)  CCAP (Coa stal Cha nge Analysis progra ms

Upload: redhousecat

Post on 09-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 1/9

This lecture reviews basic considerations whenThis lecture reviews basic considerations when

change information is extracted from digitalchange information is extracted from digital

remotely sensed data and some of theremotely sensed data and some of the

 procedures to the study of change detection. procedures to the study of change detection.

Digital Change DetectionDigital Change Detection Why do we study change detection?Why do we study change detection?

It is believed that landIt is believed that land--use/landuse/land--cover change is a major cover change is a major 

component of global change with an impact perhapscomponent of global change with an impact perhaps

greater than that of climate change.greater than that of climate change.

It is important that such changes be inventoriedIt is important that such changes be inventoried

accurately so that the physical and human processes ataccurately so that the physical and human processes atwork can be more fully understood.work can be more fully understood.

1. Classification scheme2. Geographic study area

3. Time period

4. Per pixel or object oriented approach

5. Temporal and spatial resolution

6. Phenological cycles

1. Classification scheme

2. Geographic study area

3. Time period

4. Per pixel or object oriented approach

5. Temporal and spatial resolution

6. Phenological cycles

Considerations for 

change detection

Considerations for Considerations for 

change detectionchange detection

Classifying Features

• Classification systems can be broad such as water/land, or detailed such as single family/multifamily dwelling. Thescale of the image will often dictate how finely you canclassify features.

• The USGS has developed a standard land use/land cover

classification system based in the works of Anderson 1976.

• Other classification systems have been develop by:

 – GLUT (Georgia landuse trends)

 – NARSAL ( Natural Resources Spatial Analysis Laboratory(NARSAL)

 – CCAP (Coastal Change Analysis programs

Page 2: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 2/9

Page 3: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 3/9

Page 4: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 4/9

In agricultural crops, the

analyst must be aware

of when the crops were

 planted.

A month lag in planting

date between fields of 

the same crop can causeserious change

detection error.

In agricultural crops, theIn agricultural crops, the

analyst must be awareanalyst must be aware

of when the crops wereof when the crops were

 planted. planted.

A month lag in plantingA month lag in planting

date between fields of date between fields of 

the same crop can causethe same crop can causeserious changeserious change

detection error.detection error.

Environmental Parameter: Vegetation PhenologyEnvironmental Parameter:Environmental Parameter: Vegetation PhenologyVegetation Phenology

In coastal landscapes the analyst should be aware of 

tidal differences

In coastal landscapes the analyst should be aware of In coastal landscapes the analyst should be aware of 

tidal differencestidal differences

Environmental Parameter: coastal tidesEnvironmental Parameter: coastal tidesEnvironmental Parameter: coastal tides

•Image algebra

•Post classification comparison

•Binary mask 

•An Ancillary Data Source As Date 1

•Visual on screen digitization

••Image algebraImage algebra

••Post classification comparisonPost classification comparison••Binary mask Binary mask 

••An Ancillary Data Source As Date 1An Ancillary Data Source As Date 1

••Visual on screen digitizationVisual on screen digitization

Options for change detectionOptions for change detectionOptions for change detection

Image differencing involves subtracting the imagery of 

one date from that of another.

It is possible to identify the amount of change between

two rectified images by band ratioing or image

differencing .

Image differencing involves subtracting the imagery of Image differencing involves subtracting the imagery of 

one date from that of another.one date from that of another.

It is possible to identify the amount of change betweenIt is possible to identify the amount of change between

two rectified images bytwo rectified images by band ratioing band ratioing or or imageimage

differencing differencing ..

Image Algebra Change DetectionImage Algebra Change DetectionImage Algebra Change Detection

Page 5: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 5/9

Page 6: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 6/9

PostPost--classification Comparison Change Detectionclassification Comparison Change Detection PostPost--classificationclassification

ComparisonComparison

Change DetectionChange Detection

Post-classification

Comparison

Change Detection

PostPost--classificationclassification

ComparisonComparison

Change DetectionChange Detection

Post-Classification Comparison

Change Detection: Change Matrix

PostPost--Classification ComparisonClassification Comparison

Change Detection: Change MatrixChange Detection: Change Matrix

Water Forest Urban

Water 

Forest

Urban

1 2 3

4 5 6

7 8 9

From:

2000

To:

2007

Page 7: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 7/9

Steps

1. A traditional classification of Date 1 is performed

2. one of the bands (e.g., band 3) from both dates of imagery is

analyzed using image differencing to identify areas of 

“change” and “no change” in the new image.

3. The change image is then recoded into a binary mask file

consisting of areas that have changed between the two dates.

4. The change mask is then overlaid onto Date 2 of the analysisand only those pixels that were detected as having changed

are classified in the Date 2 imagery.

StepsSteps

1.1. A traditional classification of Date 1 is performedA traditional classification of Date 1 is performed

2.2. one of the bands (e.g., band 3) from both dates of imagery isone of the bands (e.g., band 3) from both dates of imagery is

analyzed using image differencing to identify areas of analyzed using image differencing to identify areas of 

““changechange”” andand ““no changeno change”” in the new image.in the new image.

3.3. The change image is then recoded into a binary mask fileThe change image is then recoded into a binary mask file

consisting of areas that have changed between the two dates.consisting of areas that have changed between the two dates.

4.4. The change mask is then overlaid onto Date 2 of the analysisThe change mask is then overlaid onto Date 2 of the analysisand only those pixels that were detected as having changedand only those pixels that were detected as having changed

are classified in the Date 2 imagery.are classified in the Date 2 imagery.

Binary Change Mask Applied to Date 2Binary Change Mask Applied to Date 2Binary Change Mask Applied to Date 2Change Detection Using A Binary Change Mask Applied to Date 2Change Detection Using A Binary Change Mask Applied to Date 2

Instead of using a remotely sensed image as Date 1 it is

 possible to use a digital land cover map of the region.

For example, the U.S. Fish and Wildlife Service conducted a

  National Wetland Inventory (NWI) of the United States at

1:24,000 scale.

Date 2 of the analysis is classified and then compared on a

  pixel-by-pixel basis with the Date 1 information using post-

classification comparison methods.

Using An Ancillary Data Source As Date 1Using An Ancillary Data Source As Date 1 Change Detection Using An Ancillary Data Source As Date 1Change Detection Using An Ancillary Data Source As Date 1

Page 8: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 8/9

Visual OnVisual On--screen Change Detection and Digitizationscreen Change Detection and Digitization

In this process both digitized photographs (or images)

are displayed at the same time, side by side.

Both dates of aerial photography (or other type of 

remote sensor data) are visually interpreted and

compared to detect change.

Changes are digitized on screen

In this process both digitized photographs (or images)In this process both digitized photographs (or images)

are displayed at the same time, side by side.are displayed at the same time, side by side.

Both dates of aerial photography (or other type of Both dates of aerial photography (or other type of 

remote sensor data) are visually interpreted andremote sensor data) are visually interpreted and

compared to detect change.compared to detect change.

Changes are digitized on screenChanges are digitized on screen

A white arrow indicates

the direction of houses

removed from their 

foundations.

Areas of beach erosion

are depicted as black 

lines.

Areas of beach accretion

caused by HurricaneHugo are shown as

dashed black lines.

A white arrow indicatesA white arrow indicates

the direction of housesthe direction of houses

removed from their removed from their 

foundations.foundations.

Areas of beach erosionAreas of beach erosion

are depicted as black are depicted as black 

lines.lines.

Areas of beach accretionAreas of beach accretion

caused by Hurricanecaused by HurricaneHugo are shown asHugo are shown as

dashed black lines.dashed black lines.

Date 1

Date 2

Page 9: Change Dec Ti On

8/8/2019 Change Dec Ti On

http://slidepdf.com/reader/full/change-dec-ti-on 9/9

Land-use Maps