UAS and Pix4D

• Look at Step 1 (before starting a project). What is the overlap needed for Pix4D to process imagery?

Recommended frontal overlap is 75% and 60% side overlap.

• What if the user is flying over sand/snow, or uniform fields?

The recommended frontal overlap is 85% and side overlap is 70%.

• What is Rapid Check?

Faster low resolution image of the data to determine how good the dataset is.

• Can Pix4D process multiple flights? What does the pilot need to maintain if so?

Yes multiple flights can be processed.  The pilot needs to maintain a steady height so the imagery matches up.

• Can Pix4D process oblique images? What type of data do you need if so?

Yes oblique images can be processed.  The data needed is terrestrial, nadir, or aerial oblique images.

• Are GCPs necessary for Pix4D? When are they highly recommended?

They are not necessary, but they are highly recommended when processing an image with no geolocation.

• What is the quality report?

The quality report is a file that shows all of the errors and how successful the processing was.  In this case it was orthomosaic and DSM imaging.

Introduction

Pix 4D was a very simple software to use.  With that being said there are many different things that can be done using this software.  Geocoded points were captured in the aerial photos.  In this lab the students created an orthomosaic.  Some of the other features this software has are creating a flyby video showing the mapped area in 3D in an aerial view.  This software also has the option to calculate volumes (Figure 1).  This can be very helpful in mapping sand piles at mines.  Once the images are created there needs to be enough overlap so that the images can be processed (Figure 2).  Once there is enough overlap the images can be processed.  The software was used to make a 3D map of a sand mine.

Figure 1:  This image shows a sand pile with a calculated volume.  The calculated volume is 1227.16 cubic meters of sand.  This also accounts for error adding a value of plus or minus 17.88 cubic meters.

Figure 2:  This image is a snapshot from the Pix 4D online tutorial.  This image shows when there is enough overlap in the points to run the data set.

Methods

The images that were given to the class were many different images of a sand mine.  There are 68 different photos taken.  This amount ensures that there is enough overlap for this software to work.  In the first steps the class ran an initial processing tool and received a quality report (Figure 3).  The next thing we ran was a point cloud and mesh.  This allowed the class to change output options.  It can increase density of the 3D points in the map.  The final thing we ran was DSM, Orthomosaic, and Index (Figure 4).  This allows all of the images to be tied together and a 3D image to come out as a result.  In calculating the volume of the sand pile, points were marked around the outside.  Since this data has a height value, the volume can be determined (Refer to figure 1).   Also with this software a polyline can be calculated.  A poly line is a line of distance measuring something on the map.  In this blog, the student measured the length of one of the trucks in the map (Image 5).  The distance of the truck was around 12 feet.  In the data set, the polyline length was just under 4 meters. 

Figure 3:  This image shows the first page of the quality report.  This report tells the user what worked well and what didn't.

Figure 4:  This image shows the steps taken by creating the 3D map.

Figure 5:  This image shows the polyline that was collected.  The student wanted to measure the length of the truck.  There is some error as seen in the image, but it was fairly close to what is to be expected.

Conclusion

This assignment was used to get the students a background in Pix4D.  This process was much simpler than doing some of the topography assignments that were done in the past.  Now that the students have a background in this software they can use this as they get jobs in the future.  This type of mapping makes measuring topography much easier and it also goes much faster. 

GPS Topographic Survey

INTRODUCTION
In this assignment the class used a survey grade GPS to mark points in one small area on the University of Wisconsin-Eau Claire.  The GPS unit that was used, is accurate to a sub-meter.  The certain GPS that was used for this assignment was even accurate to within one millimeter.  This means that the points gathered were extremely accurate.  The points gathered were on a small hill giving the points very different elevation values.  This data for each point was placed into a text file which the students then imported into ArcGIS.  When the table was moved into ArcMap the students then made points from the XY data in the text file.  With these points the interpolation methods were then used to show the overall elevation of the location.  The interpolation methods that were used were Spline, Nearest Neighbor, Kriging, IDW, and a TIN was created (Figure 1).  (For definitions of these methods, please refer to the sandbox survey Blog post).

Figure 1:  This map shows the final product of all maps in one.  This shows the differences in the maps.  

METHODS

The data collection used an extremely accurate survey grade GPS unit, although the class ran into some minor difficulties about 20 points were still collected.  The way this area was surveyed was a random sample.  Students took the GPS unit and took points from where ever they saw fit.  This would be the best way to take a survey with nearly 20 people.  Everyone can go where ever they like and take a random point.  This also gives the students control of where they think more points should be collected.  The survey would become more accurate if more points were taken on the sides of the hill slopes.  The survey data was put into a text file.  From the text file the data was copied into excel to make a file that could be imported into ArcMap.  After the excel file was imported into arcMap the points were created using XYZ data where longitude is the X value, Y is the latitude value, and Z is the height or elevation value.  Then the interpolation methods were run to come up with the final maps.

RESULTS

According to the results more points should have been gathered around the steep sides of the slope.  In the results of this map the Natural Neighbor map came out the best and most accurate (Figure 2).  This is because of the smoothing that occurs between the different points.  This makes the map smoother and flow better.  There is much more flow in the Natural Neighbor than there is in the Spline (Figure 3) and also with the IDW (Figure 4).  The Tin image (Figure 5) shows very rough cut edges and no smoothness almost resembling a mountain.  This is obviously not accurate in this situation in the middle of the UWEC campus.  The Kriging is probably the second most accurate in this situation.  The only thing is that the hill comes to a crest and seems on the south western side to never slope back down to the sidewalk (Figure 6).

Figure 2:  This map shows the Natural Neighbor interpolation method.  This is the most accurate of the maps created.  It shows very smooth moving from point to point.

Figure 3:  This map shows the Spline method of interpolation.  This method showed the very large white portion meaning the elevation was highest here.  This map has the most white in any of the images.

Figure 4:  The IDW method shows precisely the opposite of the spline.  This map shows the smallest area of white.  This means that when the ground is only this high for a very small area.

Figure 5:  The TIN image shows the very rough areas.  This image abruptly goes from one elevation to another.  This image almost resembles a mountain.

Figure 6:  The Kriging Method shows the second most accurate of the maps created.  This is easy to see that the map very slowly and fluidly increases in elevation from the northeast to the southwest corners of the map.

CONCLUSION

These methods were the same ones that were used in the sandbox survey with different results.  This is most likely due to the fact that different survey methods were used.  The spline method produced the best looking map in the sandbox survey.  With the sandbox survey a very uniform method of sampling was used.  A point was taken at every cross in the grid pattern.  Where as in the GPS topographic survey a random sample was taken.  This could also be partly due to the fact that the class was working as one in the GPS survey rather than small groups as they were with the sandbox survey.  These maps could be much better if a stratified or a systematic sample.