Processing UAS Data

Pix4D Review

Overview:

This lab will use Pix4D software to construct a orthomosaic image. Previously, this class had only made georeferenced mosaic imagery. The software Pix4D is the current premier software for constructing point clouds, and is also very easy to use.

Before starting Pix4d, it is important to make sure the images are highly overlapped to create a 3D model that is accurate. The more overlap between images, the more accurate the 3D model will be. More overlap leads to better automatic aerial triangulation which creates a sharper 3D model. If the user if flying over sand or snow the overlap must be at least 85% frontal overlap and at least 70% side overlap. A large percentage of overlap is needed because sand and snow have very little visual content, so each overlapping image can get as much contrast between images as possible. Rapid check is to verify the proper areas and coverage of the data collection. Rapid check processes the data very quickly, but the results have fairly low accuracy.

Pix4d can process multiple flights at once as long as the coordinate system (both horizontal and vertical) of the images is the same. Oblique images can be processed in Pix4d as long as they have good overlap and GCPs. GCPs are not necessary to use Pix4d, but they are highly recommended because they create a much more accurate model. The quality report is used to find the strength and quality of the matches. 

Pix4D Software:  

Dr. Hupy provided the class with UAV imagery from a sand mine south of Eau Claire in order to complete this lab. To start, all images are imported into Pix4d mapper. The area of interest (AOI) is chosen and the flight path can then be visualized. For this lab, a freely drawn polygon was used to create the AOI. After processing the images, the quality report is then created and provides specific details about the images. Figure 1 is the quality report for this lab.

Figure 1: Summary of the quality report

Image 2 is a orthomosaic and the corresponding sparse Digital Surface Model (DSM) before densification created in the report.

Figure 2: Orthomosaic and DSM based off the report

Figure 3 is an image produced by the quality report showing areas of overlap with the images. Areas in green are the areas that have multiple overlapping images. Areas that are red and yellow are areas where overlap is poor. Areas in the middle of the image have more overlap than the edges of the image. As long as the area of interest is an area of high overlap, the output will be of high quality. 

Figure 3: The areas of overlap in the AOI

Final Overview:

This lab introduced how to quickly and easily use pix4D to process UAV data. Pix4D is a great way to visualize 3D data and produce a high quality map. Pix4D can be used by anyone with UAV data to create a map. Overall, this was a great lab to complete to finish class.

Topographic Survey

Introduction:

This lab is intended to teach how to engage in a survey of various point features on campus using a high precision GPS unit. The data will be collected as a  collectively . Each person will take turn with a partner to take a GPS point with the GPS unit. The data gathered should then be turned into continuous interpolated maps. The following interpolation methods should be used: IDW, Kriging, Natural Neighbor, Spline, and TIN.

Study Area:

The study area for this lab is a green patch of grass near the 'Sprites' between the buildings Centennial and Schofield. Figure 1 below is a map of the study area. The data points are seen below between the two academic building. This is a common area where students gather and 'chill' before or after classes.

Figure 1: A map of the study area where the data points were collected

Methods:

The data points were gathered with a survey grade GPS that has sub centimeter accuracy. The GPS records the points though a bluetooth connection. The points were gathered using a random sampling method. The random sampling method is a great method to use because it is an unbiased way to acquire a random sampling of data points. Figure 2 below are the data points that were gathered in the lab.

Figure 2: The data points gathered with the survey grade GPS

The next part of the lab is to use different interpolation methods on the acquired data points.
The following methods are: the following: IDW, Kriging, Natural Neighbor, Spline, and TIN.The inverse distance weighted (IDW) interpolation method is used to predict the elevation of the continuous surface surrounding the data points. The Kriging method generates an estimated elevation of the surfaces surrounding the data points by using the elevation of the data gathered as a reference. The Natural Neighbor method is similar to the methods above except the elevation data used as a reference is taken from data points that are near the area in question. The spline method uses a polynomial algorithm to create the continuous surface model with the elevation points gathered with the class. The last interpolation method used was TIN. A triangulated irregular network (TIN) is a representation of elevations created with triangles calculated with the gathered three-dimensional coordinates.

Results/Discussion:

The first interpolation is seen in figure 3 below. The inverse distance weighted interpolation (IDW) method shows the elevation for a small grass area with a knoll on it. The areas in the left upper portion of the map are areas of high elevation because the points were taken on the knoll above the rest of the grassy area.

Figure 3: The IDW interpolation was used on the data points in the map above

 Figure 4 seen below is a map of the data points using the kriging method. The kriging method shows the elevation for a small grass area with a knoll on it. The higher elevation areas are white or pink in color. The elevation is high in those areas because of the grassy knoll.

Figure 4: The kriging interpolation method was used on the data points in the map above

Figure 5 below is a natural neighbor interpolation map that shows the elevation for a small grass area with a knoll on it. The areas in the left upper portion of the knoll are areas of high elevation because the points were taken on the knoll above the rest of the grassy area.

Figure 5: The natural neighbor interpolation method was used on the data points in the map above

Figure 6 seen below is a map of the data points using the spline method. The spline method shows the elevation for a small grass area with a knoll on it. The higher elevation areas are white or pink in color. The elevation is high in those areas because of the grassy knoll.

Figure 6: The spline interpolation method was used on the data points in the map above

Figure 7 seen below is a map of the data points using the TIN interpolation method. The TIN method shows the elevation using a number of triangles put together creating the small grass area with a knoll on it. The higher elevation areas are the areas with red and orange coloring. The elevation is high in those areas because of the grassy knoll.

Figure 7: The spline interpolation method was used on the data points in the map above

A few weeks ago, the lab was to create interpolation maps of stratified sampling. This week, we used random sampling. The stratified sampling created a much more realistic representation of the area being surveyed versus the random sampling we just did in this lab. The random sampling interpolations created maps that were not very specific to the terrain we gathered GPS data points on.

Conclusion:

Upon the conclusion of this lab, it was clear that the stratified sampling method is a better data gathering method than then random sampling method. The stratified sampling method was used in a previous lab for this class. If this lab were to be conducted again in the future, a word of advice would be to spread out the data points in the area of interest. Another word of advice would be to make the interpolated maps somewhat translucent to see the area in which the data point was taken. Overall, this lab was interesting and informative which made the interpolation process enjoyable.