What are orthoplanes? Solar panels inspection example

An orthplane is an orthomosaic (or orthorectified image) from a facade of a building or any other vertical or angled object that has a more or less planar surface, examples: roofs, solar panels, bridge pillar,…

And what is an orthomosaic?
An orthomosaic is an image that has been corrected for optical distortions from the sensor system, apparent changes in the position of ground objects caused by the perspective of the sensor view angle and ground terrain.

In photogrammetry, an orthomosaic is created using dozen if not hundreds of pictures taken close to the subject to attain resolutions between 5cm/pixel down to 0.1cm/pixel. Pictures are then linked together by the software via keypoints, plus sensor data optimization, precise picture location and orientation determination.
Next, keypoints are used to reconstruct the scanned object in 3D in a Point Cloud. From that Point Cloud, a 3D surface can be computed such as the interactive Mesh below.
Deriving from the previous steps, the Digital Surface Model and the Orthomosaic can finally be constructed

The two main advantages of orthorectifed image are: high resolution and correction for optical distortion.
This means that an orthorectified image can be used in a GIS software and accurately overlaid with other data layers. Furthermore, accurate measurements can be performed on the image.

In contrast, if only one conventional picture was used, it would have to be taken high above the terrain so that everything were in frame, hence drastically limiting the resolution. And no accurate measurements can be performed because of distortion

Conventional picture
Click for full resolution download
Conventional picture closeup
Orthoplane closeup

Railway bridge inspection

Drone inspection of the Pont de Bory railway bridge using DJI Phantom4Pro and Pix4D mapper for photogrammetry reconstruction. Mean resolution: 0.2 cm/pixel

Bridge was built in 1925 and is still in use by the SBB/CFF train company.


Orthoplane of a side
Click for full resolution download
Orthoplane of a pillar
Click to download full resolution

Want to learn more about orthoplane? Follow this link

High resolution communal area mapping

This high resolution orthomosaic was created using a DJI Phantom4Pro drone flying between 140 and 150 meters above ground level (AGL) over about 1.6 square km.

All 800 images were then processed with Pix4D mapper to produce a 5cm/pixel resolution orthomosaic of a neighborhood in Lutry

This map can be useful for communes whishing to have a most up to date map of the built environment, rooftops available potential for solar energy, road tarmac quality can also be evaluated, built environment evolution with time, and much more

This is also followed by a 3D model of the area which could for example allow to calculate various volumes, or to even further to evaluate rooftops sun exposition for solar energy ,…

Learn more about orthomosaics

3D Point cloud

Digital Surface Model (DSM)

30 kilometres of Côte d’Azur beach virtual tour

How about taking a tour to the sunny South of France beaches and littoral path?
We have created the perfect solution for you with this 30 km virtual tour!
Put your swimming suit on, some sun cream and hop in!

This tour was created using a Freedom explorer 360° rig with six GoPro4, pictures were then assembled in AutoPano Giga, improved and corrected in Photoshop, and finally assembled in Panotour Pro. Contains 109 pictures and took a month of work to complete

Château de Brégançon – Virtual tour

We are proud to present our very first virtual tour!
This tour allows you to navigate inside the Château de Brégançon cellar and selling space.
It was captured using a Freedom explorer 360 rig with six GoPro4, pictures were then assembled in AutoPano Giga, improved and corrected in Photoshop, and finally assembled in Panotour Pro

Drone Photogrammetry – Getting accurate data


With drone photogrammetry becoming more and more popular and easier to perform, a few important things must be kept in mind when producing 3D models, especially if the requested output is used for high accuracy measurements.
When working solely with the drone’s internal GPS (and not RTK system), images are georeferenced with an error ranging between 1 to 4m to the true position. The error is due to deviations of the signal in the atmosphere, the quality of the receiver’s internal clock, number of satellites, signal multipath, and orbit errors. Fortunately this error is more or less systematic for a given time and location, all the pictures for a specific scan have about the same shift.
This means that the outputs such as the orthophoto will be shifted with the mean error between all the geotags. For the DEM however, things are slightly more complicated, as without the use of controls points systematic calculations errors can appear during reconstruction. More on that later.

If the orthophoto is used in an arbitrary coordinate system, or if no accurate comparison with other georeferenced data (this includes later scan of the same area) are planned, then things are perfectly fine and no further processing is required.
Otherwise,  introduction of ground control points, or usage of a drone equipped with RTK system is mandatory

A note on the difference between GPS accuracy and precision:

  • GPS accuracy:   The accuracy refers to the degree of closeness the indicated readings are to the actual position.
  • GPS precision:   Is the degree to which the readings can be made. The smaller the circle of unknown the higher the precision.
Difference between accuracy and precision. (www.radio-electronics.com)

Model resolution and accuracy

Inside a model you must distinguish between resolution and accuracy.
Resolution is only dependent on the flight height above the ground and camera parameters (sensor size and focal length) and is generally measured in cm per pixel. Usually the resolution is between 1 to 3 cm/pixel with consumer drones flying between 30 and 50m above ground
Accuracy however depends on the flight height, quality of the camera (lens deformation, rolling shutter,…) the type of surface surveyed, and flight planning.
Obtaining information about the accuracy of the model is only possible by having validation points. What is usually done is to evaluate the quality of a camera and optimal flight planning before field acquisition, as putting validation points during each field campaign is nearly impossible and only done for projects that require the highest accuracy.

Control points and validation points

Classic photogrammetry errors

Now what are these systematic calculations errors introduced by the software and how can careful flight planning improve the quality of the results?

Flying a drone around taking pictures in a gridded pattern and then launching the processing without any user-inputs will result in the introduction of typical systematic photogrammetry errors such as doming and tilt
With a single gridded flight pattern, images are acquired in a linear pattern along a flight line forming ‘strips’, parallel flight lines form overlapping ‘strips’, which then form ‘image blocks’ of survey areas. If systematic errors persist from strips to the block, then resulting DEM will contain the same deformation (James and Robson, 2014)
These systematic errors will be especially severe when using consumer grade equipment with wide angle – high radial deformation lenses, with the addition of self-calibration procedures

Illustration of the doming effect persisting from a single strip to a block (James and Robson, 2014)
DJI Phantom3 model of a flat football field, perfectly illustrating the doming deformation

To correct for such deformations, conjunction of two techniques is required: Ground Control Points (GCPs) and a more advancedflight pattern

Ground Control Points consist of specific target points measured with great accuracy (<2cm), which must be well visible on the drone images for later marking in the processing software.
Drone internal GPS systems are not very accurate and may result in an offset of ±2m in the horizontal axis and ±5m in the vertical axis. Consequently the inclusion of Ground Control Points allows to correctly reference the resulting models, thus permitting accurate comparisons with other georeferenced data.

Adding targets to improve and georeference your models

Testing showed that best results are acquired with a more complex flight pattern. This advanced flight planning requires to first perform a gridded flight over the entire survey area with the camera looking fully downward. Once done, capture oblique pictures with the camera at around 45° capturing the survey area at various angles, this should be performed in manual flight mode

Effect of adding oblique pictures. A comparison with dGPS data

Contact us for more details and consulting expertise

Orthoplane with Pix4D

Orthoplane is an orthorectified image of any planar object. Orthoplane can be extracted from building facades or such as in this case vertical rockfaces with the help of drone imagery
This type of rectified rockface images can be extremely useful for geologists, geomorphologists, risk assessment managers,… as it allows high resolution visualization to locations that are usually difficult to reach

For this model, a DJi Phantom3 drone and Pix4D software were used to scan a quartzite rockface over Evolène, Valais, Switzerland

Click to download full resolution image

Contact us for more details