Lab #6- Using the Gems software to construct geotiffs, and to field check your Gems data.


    Lab 6 was processing the flight mission and data from our field activity five around the pavilion on the Bollinger soccer fields. To process the images that we took, we used the Gems software. Gems stands for geo-localization and mosaicking system. Basically this software can weave all of the images that we gathered together in a form that makes sense to the human eye allowing for us to see one big picture of the study area. When processing the images the software uses two types of mosaics, fast and fine mosaics. Fast mosaics throws the images acquired down as fast as it can given the predicted alignment based on the navigation data from the sensor payload. Fine mosaics perform additional computer vision image processing techniques to finely align the imagery which takes longer to process. When possible you want to use a fine mosaic, which allows for a better image of the study area. Along with the fast and fine mosaics, the software also produces an array of different color schemes for your images. There were five different color schemes we used in this lab: RGB-fine, NDVI-FC1, NDVI-FC2, NDVI-mono, and Mono-fine. RGB stands for red, green, blue colors. Basically how the human eye would view and image in color. There are two different types of NDVI imagery, this imagery is for viewing vegetation. FC1-colors is when the redder the area the more water that is being emitted. Since we are viewing grass and pavement, we want the area were the grass is to be redder which would show that it is healthier for the grass. FC2-colors is just opposite. The greener the area the healthier and more water that is being emitted. To me personally the FC2-colors makes more sense in this case because healthy grass is naturally green. We also use two schemes of the mono-colors. Mono-colors are black and white with the white area being the healthier. This stuff that I am discussing only begins to scratch the surface of what the Gems software can do. Since it is our first lab I am looking forward to gaining a better understanding on what this software can really do.
    To begin my work in the Gems software I first had to upload our mission plan to the software. Once I uploaded the mission, I could now run the NDVI initialization. The NDVI initialization gives me my two different FC1 and 2 color schemes, along with my NDVI-mono pictures. Since our original images were taken in mono and RGB from our platforms, these image files are already in our files. After we have ran our NDVI initialization and have gotten our files, we can now generate our mosaics. Generating our mosaics will give us two different files, and overview of the files along with a tif file. What we are really wanting here is the tif files. This will allow for us to upload these files into ArcMap to create the maps we so please. A tif file originally brought up in photoshop, but can also be uploaded into ArcMap. A tif tile would look like this.

One important step when creating your running your mosaics is to check the fine alignment box along with NDVI and default color map boxes. This will give you better mosaics.

    The final step before creating your maps is to export your data to Pix4D.  Pix4D gives you the lat/long and altitude of each image. This will allow you to stitch together all the images onto a satellite base-map.

    Now comes the fun part of creating your maps for your study area. You can make a different number of maps for this depending on what you are trying to show. For the maps I created I wanted to show the different NDVI maps, RGB, and the mono maps. Along with these five different maps I also wanted to include a map of the study area without the Gems software on it.

    The first map I created was the RGB-Fine map. This map is a simple map of the study area. When I laid my RGB-tif on top of it the image was much clearer and easier to see the definition of the shed in the center. One problem that I had in the beginning with making all my maps was that the tifs had a white area around the outside of the study area which made it hard to match up with the base maps. This was a simple fix by using the mask tool in the toolbox of ArcMap. Once I created one mask for the RGB map, I could use it over again on each of the maps.

    The next map I created was the NDVI-FC1 map. This was the map with the red area emitting more water allowing for it to be healthier. In my map it came across as orange showing that it isn't emitting enough water to make it that healthy, but yet healthy enough to sustain green grass. This is the power of NDVI imaging. Although it may appear to be healthy green grass to the human eye, when you use the NDVI sensor you can see things differently allowing for you to get a different perspective on vegetation. My guess why it wasn't portrayed as red was because we conducted this mission in the fall time when the grass was entering its dormant stages.

    My next map was of the NDVI-FC2 tif. This was the same image as the FC1 image but the color scheme was different. For this map, the green area was the healthier and emitted more water, while the red areas were pavement and the small yellow area in the red was rock beds. You can see how different the FC1 and FC2 images are form each other, but in the end its just a different color scheme. Earlier I stated that the FC2 colors make more sense to me because grass is naturally green, and in FC2 the green area is the grass. This could change say if your looking at vegetation of a lake or a different scenario.

    The third map I created was the NDVI-Mono map. This map did not use fine mosaic, but instead used fast mosaic. You will be able to tell the difference between the two when there next to each other. You can also see how they are stitched together slightly different. In the mono color scheme, the white areas are the grass area while the black is the pavement and cement.

    The final map I created was the Mono-Fine map. Like I just talked about, this one did use the fine mosaic allowing for a better pitcher. Again, the white areas are the grass while the dark areas are the pavement or the building in the middle. You can easily see the difference in the stitching techniques when comparing these two maps together.

    The final map I put on here is of the study area itself. I wanted to include this as a reference so you can see what the area looked like before laying my different tifs over the area. I feel that it is important to have a reference of the study area included for this purpose.

    My final outcome would be of the six maps. One of just the study area, along with my five different maps that I created using the Gems software and ArcMap. This final map just shows the different maps that I created to show what they all look like. I could have created a number of different maps for whatever purpose I wanted to show.

    I was very new to the Gems software as this was the first time that I have ever used it when computing this data. I have only begun to scratch the surface on what the software can actually do and I feel that this software is very useful. By being able to compute mission plans and stitch together photos, this software can do just about whatever you want with the data.

    I really am interested in the NDVI applications of the software, along with showing healthy versus non-healthy vegetation. This summer I worked for a country club and talked with my boss about this course I was taking. I had already had a grasp on what NDVI imaging could do and was trying to think of ways to gear it towards a golf course. After seeing how this software can compute data from mission plans and put together images, I feel that this software could do a lot for the golfing world. That is only one small area that this software is very unique in, there are plenty of other things that it can do. I really am looking forward to learning more about the software and what it can do and give it a thumbs up from my standpoint. 








Field Activity #5- Obliques for 3D Model Construction

    In our fifth field activity we carried out a mission using oblique imagery for 3D model construction. This is the first time that we have used oblique imagery in the field. In all of our other field activities we have used nadir imagery. Nadir imagery is when the camera is pointed directly at the ground. This will give us an direct overhead view of our area of interest. Oblique imagery is slightly different than nadir imagery. Oblique imagery is when part of the horizon is in the picture as well, typically close to a 45 degree angle allowing for a side view of the area you are trying to take images of. Oblique images are perfect for allowing us to see the side view of the AOI allowing for us to compute the images into creating a 3D imagery model.
    The study area for our fifth field activity was on the soccer fields like many of our previous field activities. We took oblique images of the shed at the soccer fields using both the Iris and Gem platforms. It was a perfect day for flying with no clouds and barely any wind that made the flag move.

    In this field activity we used two different platforms. For our first mission we used the Iris platform with a GoPro camera mounted on it. The GoPro camera doesn't have GPS attached to it, so we would have to use GCP's while transferring and tying down our data when making a map of the shed. While using the Iris platform we used the structure scan mode in mission planner on the tablet. This allowed for us to set our parameters for the mission. We set the mission to oblique images and started the picture taking at 15 meters with 4 meter intervals. What I mean by this is that the mission was conducted in a cork screw fashion as it flew around the outside of shed it would raise 4 meters every time around. The images went up to 26 meters high and the images were taken at an 2 second camera photo interval. After the platform reached the 26 meter mark, it would then use cross hatching to get every nook in the roof. Upon completing this mission, we broke out the Gem platform. We did not use a mission planner for this platform, but instead flew this one manually. This allowed for us to start at a lower height and adjust the oblique angle for gain better detail on the lower section of the building along with the roof. We switched off flying around the building so everyone would get the opportunity to take pictures along with flying a multi rotor platform. 

    We have not yet learned how to process this imagery as this will be covered in future labs as the weather turns cold. When thinking about the difference between oblique and nadir imagery and the different purposes, I feel that oblique imagery has more of an upside then nadir especially with this field activity. With constructing a 3D model of a building this can pose many benefits compared to just having an overhead view like nadir imagery. This field activity was the first time that we used oblique imagery, but I look forward to seeing the differences once we have processed this imagery compared to using nadir imagery. 

    

    

Field Activity #4- Gathering Ground Control Points using various GPS Devices


     In our fifth field activity we were introduced to Ground Control Points (GCP's). GCP's are used to improve the quality of your aerial imagery acquisitions. When gathered properly they can produce data with sub-meter, and even millimeter accuracy. Likewise, when gathering your GCP's, if you do not gather them properly they can then diminish the quality of your accuracy. Another reason we want to use GCP's is that we are then able to tie down our imagery to a given coordinate system if the digital sensor doesn't have GPS on it (GoPro, Canon S110). It is important to have a coordinate system when displaying our imagery, especially for survey quality data.
     When looking at our survey area it is important to place our GCP's in a visible spot for our UAS to take images of. You don't want them placed under trees or other debris obstructing our field of vision. When placing our GCP's in our survey area it is important to have them spread out over the entire field. The closer you are to the edges of the survey field the more distorted our GCP's become. So, we do want one or two near the edges but not right on the edge while placing the other GCP's randomly throughout the rest of the survey field. Another point to bring up is when placing your GCP's with different elevation. If there is a hill in your survey area it is important to place more GCP's around and on that area. This will help with displaying the elevation of that area. It is a rule of thumb to have a minimum of three GCP's in your survey field, while it is recommended for better quality to have more. GCP's are very time consuming as you have to chart and mark where they all are. It is vital to have good field notes that you can look back on when you sketching your survey area with your GCP's. This will allow you to look back if there is any error when recording your GCP's. Pictured below is a sketch of our survey area that I took in the field while also labeling my GCP's.

    We placed six GCP's over the survey area in field activity five. We spaced them out relatively evenly throughout the survey field while making sure they weren't to close to the boundaries for distortion. Upon placing our GCP's we also recorded the GCP's with a Dual Frequency Survey Grade GPS. This GPS for us was our gold standard as it will get accuracy down to millimeters. This was the first time that we were introduced to this method of recording our GCP's so we were all relatively new to it. Some important things when dealing with this GPS was to make sure it was in the exact center of our GCP along with being level. This would allow for us to get the most accurate reading for our GCP's. Now when we recorded our GCP in the GPS it would tell us the horizontal and vertical extent along with the exact coordinates for the GCP. This would then allow us to tie down our GCP's to a coordinate system upon uploading the images. Pictured below is the gold standard GPS that we used to record our GCP's the first time. Also pictured is the first GCP that we placed in the survey field.

    The second method we used when collecting our GCP's was the Bad Elf GNSS Surveyor GPS. The GPS can produce sub-meter accuracy. This is a relatively small GPS that we lay in the center of the GCP on the ground then connect to a tablet app that will in turn allow us to record our GCP. We can also put field notes in the tablet allowing us to know the exact area for this GCP along with any other field notes that are worthy. Pictured below is the Bad Elf GNSS Surveyor that we used in the field. You can see that it is about the size of a stop watch, but is a very good device when collecting GCP data location when teamed with a tablet app.
 

    The final way we collected our GCP's was with a mobile phone. The reason we used a mobile phone GPS, is that after we analyze the data later this semester that we can then see the difference between the accuracy between the three different GPS's. We know the survey grade GPS will be the most accurate, but we want to see just how inaccurate your mobile phone GPS actually is. In today's society people rely on their mobile phones everyday when traveling and it would be good information to show and share just how inaccurate these devices are.
    The final part of our field activity was to carry out a flying mission over our GCP's and survey area. We used mission planner to plan a mission and was able to fly over our survey area. Although it was getting relatively dark when carrying out our mission, approximately 6:30 pm, we still were able to capture the GCP's over our survey field. Now that we have our mission carried out, we can now analyze the data and tie down our GCP's to a coordinate system with the three different means that we collected. We have not done this yet but I look forward to analyzing this data to see how accurate or inaccurate each means of collection was. The picture below is of our first GCP in our flight mission. This picture was taken on the first pass in our flight mission.

     When collecting our GCP's the easiest and fastest way in my opinion was with the Bad Elf GPS. All we needed was the tablet and the Bad Elf and took 10 seconds for the GCP to be collected. I also liked that we were able to put field notes right into the tablet, this could prove very helpful when analyzing the data. The second fastest method would be with your mobile phone. Just by placing your mobile phone in the middle and snapping a picture you are able to capture the GCP. Only down fall is that it is least accurate. The slowest but most effective way to capture them was with the survey grade GPS. Although it was slow and you have to be precise, you also get the most accurate GCP's. The reason GCP's are so time consuming is that you have to scan the entire survey field and determine where the proper place is to collect your GCP's. Also depending on how large or how much elevation change there is in the field. Another reason it can be time consuming is by the means you are collecting your GCP's. Are you using just your mobile phone or are you using a survey grade GPS, these will play a factor in how much time you put into collecting your GCP's. When talking about commercial and survey grade GCP's many different types come into mind. These deal a lot with permanent GCP's that need to be collected weekly, and sometimes even daily. These could deal with mines or areas that are being excavated or even areas of washout. With having permanent GCP's you know exactly where your last recordings came from and allow you to go back to that exact spot. Now when using permanent GCP's you don't need to necessarily put something in the ground. You could use a sprinkler head or another object that you know that will not change by the next time you come collect your points.
     Upon completing this field activity I was fairly educated on how to collect GCP's and the importance of them when dealing with survey grade quality mapping. Without GCP's your imagery could be well off your intended survey zone leaving you with poor quality images and mapping.