Directly after returning from Christmas I began to plan another detailed acqusition for creation of a survey-grade surface model. This project was for Century West Engineering and the site was the newly renovated Stewart Parkway in Roseburg, OR. We had been waiting for months for the roadway to be completed with light posts installed and sidewalks finished before flights. I was able to fly the 10 acre site on January 3rd in overcast and broken lighting conditions. This proved to be detrimental to the creation of the orthomosaic. Granted, the orthomosaic was never an agreed upon deliverable but as I’ve stressed about my business I like to go above and beyond what the client asks for. Because the light was variable throughout the hour long acquisition of over 1700 aerial imagees, there was a visible exposure difference between neighboring images. Traditionally I set my camera settings to all auto but in this case, with the dark pavement, vividly bright white sidewalks and intermittent sun, I ended up with a lot of imagery that was completely blown out in the whites. But.. as a testament to being prepared, I had flown a double grid pattern of imagery with 80/80 overlap so I had a ton of useable imagery. Unfortunately, it meant I had to go into lightroom and boost the imagery with an HDR effect going from frame to frame and tossing the ones that were too blown out.
Spending the time to adjust the imagery and throw out the bad frames helped produce an even more accurate surface model and even cut down on my time editing the noise out of the point cloud. I did ask our survey partner in Roseburg to head out and grab me about 7 additional ground control points along the newly completed roadway to help boost my Z accuracy which came out at an impressive 10th of a foot for the entire AOI. I then created contours for Century West at Half and quarter foot resolutions and did some final color balancing touch up of the orthomosaic in photoshop before compressing it heavily and sending it off for them.
When I was finished with the traditional Aerial Mapping surface model work I wondered if it would be possible to create a 3D point cloud by taking still frames from the 4K video I had captured doing some marketing work for them as well. To my delight I was able to make a decent looking 3D model for 80% of the roadway length, also tying it down to the right geospatial location by measuring in some aerial targets. Yes, the model is a bit noisy but it’s actually fairly good looking for pulling frames from video and just letting the software do it’s pixel matching magic!
