With every photogrammetry project there are always challenges and lessons and here at AccuPixel we like to solve problems. Along with teaching the skills for top-end photogrammetry in our Agisoft endorsed Metashape Courses we deliver specific consultancy to key clients – from Indonesian mining surveys to coral reef monitoring in Florida.
We also offer support to our customers and anyone buying training, UWIS or Metashape licenses from us qualifies.
Misalignment
One of our customers approached us for some guidance about a drone survey they had just completed. The survey was an oil & gas industrial site containing pipework, moving vehicles and standing water – all of which can cause issues in just about every output.
Whilst the overall accuracy and scaling of the industrial site was exceptional there were areas of the ortho photo that were not quite perfect. Our customer wondered if the survey itself was causing issues? Or was it the steps used to process the images for the photogrammetry project? Most important – Could the survey results be improved at all?
Of all the outputs we can generate the ortho photo is perhaps the harshest critic.
Here’s an example; In poor underwater visibility divers will often lay a thin white nylon line to help navigate a site. A quick visual check after processing reveals how well the nylon line was captured:
The above image is a snip of a larger site scanned in 2019 – a now destroyed live torpedo Simon digitally preserved in Scapa Flow and you can see the ortho photo of the whole site hosted by our friends at DroneLab right here.
The nylon line has a couple of jumps or stepped gaps. One or two gaps is not a huge concern but it can highlight issues with image overlap or the processing steps used.
Improving the Output of Photogrammetry projects
The survey site was certainly challenging. The myriad of pipes and supporting hardware was complex and containing a lot of thin structures.
And just like the thin nylon diver guideline the pipes and supports were showing a little misalignment.
It’s important to remember there are several options and choices when processing your photogrammetry project in Metashape, plus there are a lot of settings to tweak. Metashape is as much a toolbox than a single point solution and this gives us a range of options. There is more data and detail lurking in the images and Metashape can extract it and this was the first step.
The second step was to refine the ortho photo and selectively apply source images to specific areas. This process targets areas where we detect vehicle movement. This handy tool works with patches and polygons to isolate any source image(s) that have caused motion blur.
The Solution for Photogrammetry projects
In this instance, the quality of the ortho photo improved by changing its source. Instead of using the Dense Cloud, Jose skipped this stage entirely. He went straight from the optimized sparse cloud to the mesh. He used depth maps at the highest quality as the source.
The results were a finer level of detail in the smaller features and depth maps are a better route to a more refined result than using the dense cloud.
Summary
Meanwhile, Jose returned the finished the photogrammetry project to the customer along with some recommendations for data capture improvements. When capturing images, we improve success by ensuring the drone, camera, and lens match. This setup ensures they work together to capture as much detail as possible.
The next critical part is the flight planning and image capture methodology. Just like Metashape there are a huge range of options for both and this is where Jose’s vast experience with drones and aerial image capture is vital.
In conclusion, all of this is part of the service AccuPixel clients receive. Hence, this is what we do – helping others deliver consistent and high quality photogrammetry results via training & support or consultancy for those really challenging use cases.
Lastly, please use the contact us page to drop us a line and discuss your project challenges – we look forward to hearing from you.