Introduction:
 |
Figure 1: Demonstration of DSM with and without GCPs
|
In this lab we took a bit of a step back and went through the data processing from the wolfcreek data we've been working on in the previous lab from the beginning. Our main goal of this lab was to compare the difference between data collected with and without ground conrol point (GCP) corrections. We began by using Pix4D, an image processing software, to give us a 3D model of the terrain; after that we processed the data into a 3D map. Once the processing was done we were had a digital surface model (DSM) and an orthomosaic map.
Methods:
We began this lab by opening Pix4D and starting a new project. As always we organized our data with the same process we discussed in last weeks lab. We also named our new project in Pix4D by: ddmmyyy_location_aircraft/sensor_altitude_coordinate correction system. (coordinate correction system would be like ground control points). The next step is adding the images or folder path for the images we were going to use. Next we chose what coordinate correction system we would use for our data, but sense we were comparing our previous labs data with GCP correction we selected nothing for this step. Finally we could choose what to do with pour data; our ultimate goal was to get our DSM, but we also used Pix4D to generate an orthomosaic map and and a flyover animation of the map, which you can see below in figure 2.
 |
Figure 2: Orthomosaic Animation |
Pix4D processes data in 3 broad steps: 1. initial processing 2. point cloud and mesh, and 3. DSM, orthomosaic and index. Initial processing is basically a quick check of the data to weed out any issues with the data and we would run this step of the process first in order to generate a quality report of the data. Once this initial processing was done (this time it took about 5mins) Pix4D generated a quality report that we checked for data errors and could to preview how our orthomosaic and DSM would turn out. Once we were satisfied with the quality of our data we could check the other 2 boxes for our processing. Pix4D can process all 3 of the aforementioned steps all at once but this is generally not a good idea as if there are problems with the data they can mess up your final product.
Step 2. is generating point cloud and mesh data. Point cloud is a set of data points that are on the surface of whatever image is being made. The animation in figure 2. above shows the point cloud and mesh data together. Note that it is a 3D image. After this, in step 3. the DSM and orthomosaic are processed and developed. These are raster files and are the same as the one we used last week in ArcPro. The key difference in this lab is that there are no GCP corrections. As a result the elevation values are very different.
 |
Figure 3. Pix4D DSM without GCP corrections |
 |
Figure 4: Pix4D Orthomosaic image |
Results & Discussions:
Pix4D organizes it's data much like ow we organize ours. The three processing steps each have their own folder and organizes its data within those folders. For example the quality report will be in the initial folder, which contains another folder for reports. Within this report you can also find processing times for each of the 3 steps. You can see this table below in figure 5. In the point cloud folder we can wholes and deformations in the image data from a birds eye view. This is due mostly to the fact that the photos for this flight were taken from directly above thus, vertical entities such as the sides of buildings or very steep terrain can't be seen by the photo. This leaves a "whole" in the data and when you try to stitch the images together with meshing they look distorted.
 |
Figure 5: Pix4D Processing times |
Conclusion:
Pix4D is a great tool for processing image data. The initial processing step can save a lot of time and grief by checking image data quality and allowing for a preview of the processed data to be viewed. Pix4D also is what we predominately use to create our raster data such as the DSMs and orthomosaics. I Know of no other software that does this.
No comments:
Post a Comment