Balloon Mapping Part I

Introduction:

This week, my class and I launched our first balloon as a test run for our final balloon mapping exercise. We are using balloons to capture aerial images of the University of Wisconsin-Eau Claire campus which we will later mosaic to create an aerial map of the campus. This test launch proved to be critical for the success of our final launch as we were able to find and correct the issues we encountered with the balloon, rig, camera, and so on before the final launch. We also began to stitch together the images we collected using one or more of several mosaicking programs—mapknitter, ArcMap, and ERDAS. Mosaicking is simply overlaying rasters (aerial images in particular) and creating one seamless output raster out of them. The final result should appear to be a single faultless image.

Methods:

To start off the launch, we first had to fill the balloon with helium and tie it off. We used a plastic tube to fill the balloon and had a student holding the balloon to keep it from flying away (Figure 1). The balloon had a diameter of 5 ½ feet, so it grew to be very large. Once it was sufficiently filled, we placed a plastic ring around the bottom of the balloon for the carabiners and tied it off using three zip-ties. The carabiners were used to attach both the 400 feet of anchor string and the rig to the balloon.

Figure 1: This photo was taken while filling the balloon with helium. One member of the class had to hold the balloon to keep it from flying away (on the left), another (middle--me!) had to hold the bottom of the balloon to keep it sealed around the plastic tube we were using to fill it, and another had to monitor the helium tank (right).

Over the past few weeks several students in our class have been working with our professor to continue developing the camera rigs we began creating back in Week 3. They decided to use a small Styrofoam box with a hole cut out for the camera lens as our rig rather than the soda bottles (Figure 2). The digital camera was simply placed inside the Styrofoam box with the lens fitting perfectly into the hole. We turned on the continuous shot mode and taped the box shut using some masking tape. We also attached a small GPS device to the rig (Figure  2). To connect to the balloon, four strings were tied to the box and then knotted together with a carabiner that would also be attached to the balloon.

Figure 2: This photo points out several of the key components of the balloon and rig. The four strings holding the rig can also be seen in the photo even though they are not labeled.

After everything was fastened, we were ready to launch the balloon. We simply let the balloon go until the anchor string had reached 400 feet and held it at this length (this was marked off before the launch). Unfortunately, the wind was very strong that day and the balloon was blown sideways so that it only reached about 100 feet in altitude rather than the 400 we were expecting (Figure 3). Once the balloon reached its maximum height, we walked it around the campus careful to avoid trees, buildings, and other objects that could snag the string. The rig was also affected by the wind and ended up sideways for quite some time. We were able to walk around the entire campus mall and then decided to bring down the balloon and place a flip cam inside the rig to take aerial video of campus.

Figure 3: In this photo, the balloon had reached its maximum height of around 100 feet in the air. As can be seen, the wind blew it very far to the side and it was unable to reach the 400 foot height we were hoping for.

We made several adjustments to the rig before we launched the balloon again, and we also filled the balloon with more helium in hopes that the balloon would reach a higher altitude. The first adjustment we made was to secure the camera inside the rig using more than just the hole for the camera lens. With such strong winds, the camera could have easily been knocked on its side inside of the rig and taken pictures of the inside of a box for the entire launch (fortunately this did not happen). This was corrected by taping the camera to the inside of the box. We also decided to tape the four strings that attached our rig to the balloon along the sides of the Styrofoam box so as to keep the rig from turning sideways.

After we had made our modifications, we walked through the mall once again and then took the balloon across the river that runs through our campus as well. This turned out to be a poor decision on such a windy day, and our rig broke off from the balloon after some hefty gusts of wind (Video). Somehow, the rig remained intact and fell into the river rather than onto a building or on concrete. Our professor was able to fish it out with a large stick and we recovered all of the footage we had taken. Sadly, the balloon was not able to be retrieved and flew off into the sky never to be seen again.

Video: This is a video that captured the fall of our rig from the balloon into the river. In the beginning of the video, the effect of the  wind on the balloon and rig can be seen.

 Once the launch activity was complete, we were responsible for mosaicking a sample of images from the launch using a program of our choice. I chose to try out Mapknitter, an online mosaicking program because I had not yet learned how to perform a mosaic in either ArcMap or ERDAS. Mapknitter provided several basemap options with which to overlay our aerial images. I used the Google Images basemap because it seemed to be the most accurate. I then uploaded the images that I wished to use and tried to match them to the basemap and one another as best as I could. The area I chose to work with was the only area that seemed to have suitable, top-down aerial photos for mosaicking. It is important that the aerial photos be taken as perpendicular to the ground as possible so as to prevent distortion. The footage from the flip cam could not be used for mosaicking purposes so we were confined to the photos taken in the first part of our launch. There were several tools that allowed me to rotate, resize, and warp these images to my liking. The program worked fairly well for this project, but the mosaic was still imperfect due to the varying angles of the aerial images (Figure 4). This would have been an issue with any of the three programs and couldn’t really be helped.

Figure 4: This is the final sample mosaic I created of a small portion of the UWEC campus.There are some imperfections, but I was surprised to see that the mosiac worked as well as it did with the photos we had collected.

Results/Discussion:

This practice launch was a great learning experience for the class and helped to make the final launch run much more smoothly. We corrected for issues with the security of the camera inside of the rig and the rig itself during class. These adjustments helped to capture aerial images that would be more steady and better for mosaicking (Figure 6). Another lesson we took away from the practice launch was to be aware of the weather conditions. The wind really threw off our desired height for the balloon and tossed around the rig quite a bit. This resulted in some crazy images from all sorts of angles that couldn’t be used for any sort of mosaicking. Even the better images had angles that made them difficult to work with (Figure 5).

Figure 5: This is an example of a photo that is unsuitable for mosaicking images or making maps. Though it is a "pretty" picture, the angle is so great that it would be inappropriate to use for our purposes.

 

 

Figure 6: This is one of the best photos we captured from the balloon launch. It is nearly perpendicular to the ground, though we can see there is a slight angle by looking at the side of the building on the left.

 

 

 

As far as the mosaicking was concerned, I found that Mapknitter was a decent option, but I would choose to work with either ArcMap or ERDAS in the future. The major issue I had with Mapknitter was that the images couldn’t be easily stacked in the order I would have liked with the highest quality image placed on top of the others. The order of image stacking is very important while mosaicking and can completely alter the way the mosaic appears. Both ArcMap and ERDAS allow the user to set their stacking order prior to mosaicking, so I will be using one of the two for the images from our final launch.

 

Figure 7: This figure highlights an area where my mosaicked images and the basemap were greatly different. This is due to the fact that we could not get suitable photos for every part of campus.

Conclusion:

The preliminary balloon launch we performed this week was extremely instructive for our class. We were able to assess the high and low points of our rig and make the corrections necessary for a successful final launch. We also were able to work with several programs to create a mosaicked image out of the aerial photos we collected. I found that Mapknitter worked fairly well, but I will be using a different program to mosaic the images for our final launch.