Bluesky Develops Low-Cost Measurement Tool for Capturing Accurate 3D Spatial Data with SmartPhones

Aerial mapping company Bluesky of Leicestershire, UK has completed a research project backed by the UK government’s innovation agency, Innovate UK, to develop the use of mobile phones for capturing accurate 3D spatial information.

The nine-month research project focused on the use of standard smart phone technology to capture and calibrate video footage, and then convert it to 3D information. Designed for electricity Distribution Network Operators (DNO) and other organizations with a distributed asset base, the low-cost measurement tool can provide an accurate record of the feature’s location and its environment. Accurate measurements of essential infrastructure, such as overhead power lines and other utility facilities, could then be extracted using specially developed algorithms and workflows.

According to company materials, during the project, Bluesky tested a number of hardware, software and deployment options. These included the use of aerial photography to add control points to the video footage. As the project progressed, it was established that in remote areas there were insufficient features, for example road markings, lamp posts or buildings, to establish the required control. Bluesky therefore developed alternative innovative methodologies including the use of a calibration object or the measurement of a feature within the imagery.

Working alongside project partner ADAS, Bluesky also undertook rigorous testing of the solution establishing and documenting the field data capture process, identifying minimum hardware requirements, such as camera pixel capacity, and additional developments to the data delivery mechanism. Following minor enhancements and additional trials, Bluesky hopes to launch the mobile phone mapping tool, complete with data processing and hosting services, in Q2 2018.

In the following interview with GISCafe Voice, James Eddy, technical director, Bluesky, spoke about the project and how the new measurement tool will be used.

How were you able to convert the video footage to 3D information? 

We break the video down into individual frames and then perform a triangulation process, but tying the images together using tie points (the same point on each image).  We then perform a dense matching process on these images which works out the 3D location of each pixel. It is. Wry similar to what we do with the aerial photography.

Did you create a 3D model? 

No. We create a point cloud rather than an actual model.

What kinds of specially developed workflows and algorithms were created for taking the measurements of utility poles,etc?  

All of the algorithms have been specially developed for this project, but in general there is the triangulation algorithm which ties the images together and works out the orientation of each image. Then there is the dense matching algorithm which creates the point cloud. The actual measurement of the utilities is done in a cloud-based point cloud viewer which is open source.

What types of geographic data was included? 

The only geographic data used is the GPS from the phone and the images from the video. It is entirely standalone.  We have got one version which you add ground control from google earth, but the best results are when we simply scale the image using a known object within the point cloud, such as a person or a sign or pole etc. This allows the user to measure accurately within the point cloud but it is relative, ie does not have geographic coordinates.

Was aerial photography added after the video footage? 

No. We did not use aerial photo, but in the future it maybe that we look at fusing them together.

How did you come up with things like calibrating objects or measuring a feature within the images? 

We tried to use ground control using Google Earth, but it was getting complicated and the results were not as good as we wanted. So we thought if there was an object of known size in the data we could simply scale it.

How would you compare this with 3D scan data being gathered using a smartphone? 

I can’t comment on this as I have not seen such data. But from what I understand these phones that scan are very limited. This can be used in almost any environment, and just uses a normal phone, with a camera.

Are there any special features needed on the smartphone in order to create the 3D maps?

No, just a normal phone (or normal camera) will do it. However the better the phone the better the results. We found iPhones gave better results than cheaper Android phones for example.

Where do you see this project going from here?  

We are hoping to get some of the DNOs to do field trials with the system, but in the immediate term we are doing a few enhancements to it to make it easier to use.  In the longer term we see other applications, such as heritage, property, accidents recording, asset management etc. We are hopeful that it will fill a niche.

Innovate UK praised Bluesky and awarded the project a score of 4 out of 5. Bluesky now must take the proof of concept design into a market ready solution.

Tags: cloud, crowdsourcing, data, geospatial, GIS, Google, Google Maps, imagery, Infrastructure, intelligence, location, mapping, maps, mobile, satellite imagery, smartphones

Categories: analytics, Bluesky, climate change, cloud, cloud network analytics, data, geospatial, GIS, government, lidar, mapping, Open Source, photogrammetry, public safety, resilient cities, satellite imagery, smartphones, subsurface utilities, transportation, utilities

This entry was posted on Wednesday, February 28th, 2018 at 10:29 am. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.