Investigating Photogrammetric Methods for Stream Planform ... · Investigating Photogrammetric...

1
+ Investigating Photogrammetric Methods for Stream Planform Identification Michelle Lohberg 1 , Kyle Lusk 1 , David Miller 1 , Forrest Stonedahl 2* , and Susa H. Stonedahl 1* 1-St. Ambrose University 2-Centre College *Faculty Advisors I. Overview II. Methods Outlining Methods Camera GPS Method Take photos along bank edges Convert longitude/latitude to (X,Y) coordinates Plot the graph 3D Imaging Method Use 3D setting on camera or take simultaneous pictures (twinned cameras) 3D images mirror human vision, which our brains can use to recognize depth Extract depth information (future work) Cross-Sectional Method Set-up poles in-line with each other spaced ten feet apart Create a new line of poles when needed due to curvature of stream and measure the angle between lines Wrap string around poles and mark it at regular intervals Measure distance to each bank from the marked locations holding measuring tape at a 90 o angle to the string Measure the distance down stream and distance to bank of each control point. Use trigonometry to plot the outline and control points Triangulation Method Measure the distances from two poles located in the center of the stream to points on the stream outline approximately every three feet Repeat for 2 additional pairs of poles down stream Measure to a point on each side of the stream using both the upstream and downstream pair of poles from these points Use above to calculate the angle between the lines formed by each pair of poles Use trigonometry to plot the outline and control points Multi-view Picture Reconstruction Method (Photosynth) Take many pictures from many angles Upload them to Photosynth to create a point cloud of common points Extract the point cloud and rotate it to an overhead view Square Method Build a large wooden square, mark corners with a bright color, distinguish one corner from the rest Secure the square in the stream Take pictures from many angles, capturing as much of the bank as possible Process the pictures using GIMP, StreamGUI, and MATLAB (see section III) IV. Results VII. References and Acknowledgements Barker, R., Dixon, L., Hooke, J., 1998, Use of Terrestrial Photogrammetry for Monitoring and Measuring Bank Erosion, Earth Surface Processes and Landforms, 22(13):1217-1227 Bird, S., Hogan, D., Schwab, J., 2010, Photogrammetric Monitoring of Small Streams Under a Riparian Forest Canopy, Earth Surface Processes and Landforms, 35(8): 952-970, doi:10.1002/esp.2001 Chandler, J., Ashmore, P., Paola, C., Gooch, M., Varkaris, F., 2002, Monitoring River-Channel Change Using Terrestrial Oblique Digital Imagery and Automated Digital Photogrammetry, Annals of the Association of American Geographers, 92(4):631-644 Stonedahl, S.H., Harvey, J.W., Detty, J., Aubeneau, D., Packman, I., 2012, Physical Controls and Predictability of Stream Hyporheic Flow Evaluated With a Multi-Scale Model, Water Resources Research, 48(10), doi:10.1029/2011WR011582 We would like to thank Dr. Katie Trujillo, Dr. Jodi Prosise, and Vickie Logan for organizing the St. Ambrose Undergraduate Summer Research Institute and to extend a special thank you to our private donors who have made the St. Ambrose Undergraduate Summer Research Institute possible through their generous support. V. Conclusions Camera GPS It is not possible to directly use consumer grade in-camera GPS equipment to create a reasonable stream outline Cross-Sectional and Triangulation (Non-Photo) Very time consuming, tiring and disturbs sediments in the stream due to the placement of poles and walking in the stream Very accurately measured control points, but would be hard to do at a high resolution High resolution requires very specific measurements at short intervals which increases the amount of field-work Multi-view Picture Reconstruction (Photosynth) Photo clouds (a grouping of colored pixels found in multiple pictures) made it difficult to locate control points. This made it challenging to create a standard of comparison with the other methods It seems as if the point cloud does not find points within the stream often, but we do not know that it finds points right along the bank either Square Method (Non-Photo) Requires a lot of post processing in the GIMP image editing software, but was relatively easy to do in the field This method was very accurate and provides a high resolution outline Stream outlines and calculated control point locations from five of our methods along with percent error associated with each of these methods Cross-Sectional Method Twinned camera setup 8’x8’ square in stream Try and apply methods to streams both bigger and smaller than Black Hawk Creek to evaluate potential strengths and weaknesses of the outlining process Make control points more visible in areas with a mixed amount of sunlight. (Adjusting exposure was not always enough) Automate process to mark-up pictures for use in the square method’s graphical user interface. Extract the best outline coordinates from the layered visual output from the square method Further investigate the 3D/stereo imaging VI. Future Work III. Square Post-Processing Stream planforms (aka outlines) are used to map scientific measurements, estimate volumetric discharge, and model stream flow. Changes in these planforms can be used for quantifying erosion and water level fluctuations. This summer we investigated a variety of cost-effective methods of generating these planforms. We mapped the banks using the GPS feature in consumer-grade digital cameras, used twinned cameras to create 3D (stereo) images, built point-clouds using multi-view stereo software, and developed procedures for two direct measurement methods. Along with these methods we developed a novel photogrammetric procedure, which involved floating a large wooden square on the creek, photographing the square and banks from as many angles as possible and then using the square to correct the perspective of the images and extract the outline. We also developed a new open-source software application to assist with this method. We collected data for the six methods at Blackhawk Creek in Sunderbruch Park in Davenport, IA and analyzed the effectiveness of each method in terms of accuracy, environmental intrusiveness, efficiency, and ease of use. Triangulation Method Photosynth point cloud Control Points 30 control points (red circles) were placed along edges of stream We measured 88 lengths between pairs of control points Approximated error in the various stream outlining methods as the average percent error in control point lengths Control point GIMP Pictures Manually label each control point and square corner shown. Trace all visible bank StreamGUI Opens images User numbers control points User labels square corners Saves coordinates of points and outlines to *.txt file Note: We created the StreamGUI program and MATLAB scripts MATLAB 1) Corrects for radial distortion using Brown’s model. Parameters found with AGIsoft Lens 2) Corrects for perspective by assuming corner pixels of square form a parallelogram in 3D space 3) Rotates each outline and orients it on a plane with respect to its square 5) Overlaying with weighted transparency forms final outline 4) Rotates and translates in order to overlay outlines from different squares Square 1 Square 2 Square 3 Camera GPS Cross-Sectional Triangulation Photosynth Square 1.51% 2.04% 1.31% 11.9% 298% Stereo image from twinned cameras. (View with red/blue glasses for 3D effect.)

Transcript of Investigating Photogrammetric Methods for Stream Planform ... · Investigating Photogrammetric...

Page 1: Investigating Photogrammetric Methods for Stream Planform ... · Investigating Photogrammetric Methods for Stream Planform Identification Michelle Lohberg1, Kyle Lusk1, David Miller1,

𝐾 =𝜋𝐷

11+𝐿

∆𝑇ln

𝐻0

𝐻1,

Investigating Photogrammetric Methods for Stream Planform IdentificationMichelle Lohberg1, Kyle Lusk1, David Miller1, Forrest Stonedahl2*, and Susa H. Stonedahl1*

1-St. Ambrose University 2-Centre College *Faculty Advisors

I. Overview

II. Methods

Outlining Methods

Camera GPS Method

• Take photos along bank edges

• Convert longitude/latitude to (X,Y) coordinates

• Plot the graph

3D Imaging Method• Use 3D setting on camera or take simultaneous pictures (twinned cameras)

• 3D images mirror human vision, which our brains can use to recognize depth

• Extract depth information (future work)

Cross-Sectional Method• Set-up poles in-line with each other spaced ten feet apart

• Create a new line of poles when needed due to curvature of stream and

measure the angle between lines

• Wrap string around poles and mark it at regular intervals

• Measure distance to each bank from the marked locations holding measuring

tape at a 90o angle to the string

• Measure the distance down stream and distance to bank of each control point.

• Use trigonometry to plot the outline and control points

Triangulation Method• Measure the distances from two poles located in the center of the stream to

points on the stream outline approximately every three feet

• Repeat for 2 additional pairs of poles down stream

• Measure to a point on each side of the stream using both the upstream and

downstream pair of poles from these points

• Use above to calculate the angle between the lines formed by each pair of

poles

• Use trigonometry to plot the outline and control points

Multi-view Picture Reconstruction Method (Photosynth)• Take many pictures from many angles

• Upload them to Photosynth to create a point cloud of common points

• Extract the point cloud and rotate it to an overhead view

Square Method• Build a large wooden square, mark corners with a bright color, distinguish one

corner from the rest

• Secure the square in the stream

• Take pictures from many angles, capturing as much of the bank as possible

• Process the pictures using GIMP, StreamGUI, and MATLAB (see section III)

IV. Results

VII. References and Acknowledgements• Barker, R., Dixon, L., Hooke, J., 1998, Use of Terrestrial Photogrammetry for Monitoring and

Measuring Bank Erosion, Earth Surface Processes and Landforms, 22(13):1217-1227• Bird, S., Hogan, D., Schwab, J., 2010, Photogrammetric Monitoring of Small Streams Under a

Riparian Forest Canopy, Earth Surface Processes and Landforms, 35(8): 952-970, doi:10.1002/esp.2001

• Chandler, J., Ashmore, P., Paola, C., Gooch, M., Varkaris, F., 2002, Monitoring River-Channel Change Using Terrestrial Oblique Digital Imagery and Automated Digital Photogrammetry, Annals of the Association of American Geographers, 92(4):631-644

• Stonedahl, S.H., Harvey, J.W., Detty, J., Aubeneau, D., Packman, I., 2012, Physical Controls and Predictability of Stream Hyporheic Flow Evaluated With a Multi-Scale Model, Water Resources Research, 48(10), doi:10.1029/2011WR011582

We would like to thank Dr. Katie Trujillo, Dr. Jodi Prosise, and Vickie Logan for

organizing the St. Ambrose Undergraduate Summer Research Institute and to

extend a special thank you to our private donors who have made the St. Ambrose

Undergraduate Summer Research Institute possible through their generous

support.

V. ConclusionsCamera GPS • It is not possible to directly use consumer grade in-camera GPS equipment to

create a reasonable stream outline

Cross-Sectional and Triangulation (Non-Photo)• Very time consuming, tiring and disturbs sediments in the stream due to the

placement of poles and walking in the stream

• Very accurately measured control points, but would be hard to do at a high

resolution

• High resolution requires very specific measurements at short intervals which

increases the amount of field-work

Multi-view Picture Reconstruction (Photosynth)• Photo clouds (a grouping of colored pixels found in multiple pictures) made it

difficult to locate control points. This made it challenging to create a standard

of comparison with the other methods

• It seems as if the point cloud does not find points within the stream often, but

we do not know that it finds points right along the bank either

Square Method (Non-Photo)• Requires a lot of post processing in the GIMP image editing software, but was

relatively easy to do in the field

• This method was very accurate and provides a high resolution outline

Stream outlines and calculated control point locations from five of our methods along with percent errorassociated with each of these methods

Cross-Sectional Method

Twinned camera setup

8’x8’ square in stream

• Try and apply methods to streams both bigger and smaller than Black Hawk

Creek to evaluate potential strengths and weaknesses of the outlining process

• Make control points more visible in areas with a mixed amount of sunlight.

(Adjusting exposure was not always enough)

• Automate process to mark-up pictures for use in the square method’s graphical

user interface.

• Extract the best outline

coordinates from the

layered visual output from

the square method

• Further investigate the

3D/stereo imaging

VI. Future Work

III. Square Post-ProcessingStream planforms (aka outlines) are used to map scientific measurements, estimate volumetric discharge, and

model stream flow. Changes in these planforms can be used for quantifying erosion and water level fluctuations.

This summer we investigated a variety of cost-effective methods of generating these planforms. We mapped the

banks using the GPS feature in consumer-grade digital cameras, used twinned cameras to create 3D (stereo)

images, built point-clouds using multi-view stereo software, and developed procedures for two direct

measurement methods. Along with these methods we developed a novel photogrammetric procedure, which

involved floating a large wooden square on the creek, photographing the square and banks from as many angles

as possible and then using the square to correct the perspective of the images and extract the outline. We also

developed a new open-source software application to assist with this method. We collected data for the six

methods at Blackhawk Creek in Sunderbruch Park in Davenport, IA and analyzed the effectiveness of each

method in terms of accuracy, environmental intrusiveness, efficiency, and ease of use.

Triangulation Method

Photosynth point cloud

Control Points• 30 control points (red circles) were placed along edges of stream

• We measured 88 lengths between pairs of control points

• Approximated error in the various stream outlining methods as the average

percent error in control point lengths

Control point

GIMP

Pictures• Manually label each

control point and square corner shown.

• Trace all visible bank

StreamGUI• Opens images• User numbers

control points• User labels square

corners• Saves coordinates of

points and outlines to *.txt file

Note: We created the StreamGUI

program and MATLAB scripts

MATLAB

1) Corrects for radial distortion using Brown’s model. Parameters found with AGIsoft Lens

2) Corrects for perspective by assuming corner pixels of square form a parallelogram in 3D space

3) Rotates each outline and orients it on a plane with respect to its square

5) Overlaying with weighted transparency forms final outline

4) Rotates and translates in order to overlay outlines from different squares

Square 1

Square 2

Square 3

Camera GPS Cross-Sectional Triangulation Photosynth Square

1.51%2.04% 1.31% 11.9%298%

Stereo image from twinned cameras.(View with red/blue glasses for 3D effect.)