Direct Georeferencing using a GNSS/INS:What is it?

Article1: Video Transcript- Direct Georeferencing using a GNSS-Inertial Navigation System (INS): What is it?

Direct georeferencing is when you can assign a geographical location on Earth to a pixel from a camera image or a digital point from a laser WITHOUT any additional measurements referencing the ground.

To make this really simple, you can take a digital picture of a scene, and because you know the position of the image and its pointing direction relative to the ground, for every object you see in the picture, you can measure its location on Earth.

So I don’t have to send someone out into the unknown to figure out where it is. I can get it from my photos.

This concept was revolutionary for the aerial mapping industry because maps that used to take months to make could suddenly be created within days.

That’s because we could eliminate the need to send out people into the field to gather ground points, which took months. Then we eliminated or enhanced aerial triangulation method, which used to take weeks.

Now, Direct Georeferencing allows us to make high-accuracy maps within hours and for certain applications, we already have real-time, on-board processing where maps are being made while the plane is still flying.

The most common way to do direct georeferencing in airborne, land, and marine mapping is with an integrated GNSS Inertial Navigation System.

The question is how does this work?

Very simply, using the example of an aerial camera, you need to know four things:
Number ONE: You need the geographical location of the camera. For this you have, GNSS, which stands for Global Navigation Satellite System, and GPS is an example of one. You probably use the GPS device in your smart phone and tablets and even the navigator in your car. GPS gives you the geographical latitude, longitude, and even elevation of your camera relative to the earth. These are geographic coordinates, which represents an ellipsoidal or rounded earth.

Number TWO: The Inertial navigation system or (INS) gives you the pointing direction or orientation of your camera w.r.t. the ground. This means, I know my roll, pitch, AND my direction w.r.t. to North Pole. This system is based on an Inertial Measurement Unit and it is usually sitting on your sensor.

Number THREE: You make a map projection where you convert your sensor’s  geographical coordinates and orientation for each picture to a cartesian coordinate system. You want to represent a three-dimensional body, like the earth onto a plane, like a piece of paper. This means, you’re taking a rounded object, and flattening it out. So why would you do this? Very simply, when this is flattened, you can measure distances, areas, shape, direction, bearing, and scale. Adding the elevation component lets you measure heights, volumes, and slopes. This opens many possibilities.

Number FOUR: People forget this one! You HAVE to know the inside geometry of the camera or sensor. This means, you know the CCD of the camera so well, that you can create an internal coordinate system for every single pixel in it. AND, this means your equipment is engineered so well that the parts are stable to a 1000th of a degree and by a pixel size, which can be about five microns.

So Direct Georeferencing is serious precision and engineering!

Now, these concepts can be applied, not just for airborne, but for land, mobile, indoor, and marine mapping applications.

This is GeoErnest.

About GeoErnest
This is GeoErnest. Visit geoernest.com

7 Responses to Direct Georeferencing using a GNSS/INS:What is it?

  1. Pingback: Ernest Yap’s Blog on Direct Georeferencing | WiLDER LiDAR Blog

  2. Josef says:

    Ernest,

    this is a great Video and I like the way how you represent it.
    I have the following comments:
    Direct Georeferencing is working very well if one considers the rules that should be considered with direct georeferencing. But we know from experience that often people do not follow the rules and even they do so the orientation and positioning is not accurate enough for parallax free stereo viewing and image matching.
    In such cases a traditional aerial triangulation is still necessary and advisable.
    An aerial triangulation is also very helpful to detect and eliminate camera distortion problems or problems with the principal point definition.
    Most cameras should be time by time calibrated with a calibration flight and an accomplished in-situ calibration.

    Overall the aerial triangulation is quite easily applicable when GNSS/IMU data is available and with a bit more effort an independent control of the block can be applied to further improve the orientation data to control the orientation and to control the camera.

    Josef

  3. GeoErnest says:

    Josef, what you wrote is absolutely correct! Direct georeferencing (DG) is not for all cases!
    I have a follow-up video that addresses the “rules” of DG for next week. As you know, this is an on-going education process, and I am glad that your comment will help us do this. You have given me an awesome video idea about aerial triangulation (AT).

    I also encourage users of direct georeferencing to own AT software to perform assisted-AT for the reasons you gave. They can use assisted-AT as standard workflow or as a powerful backup. The processing time and the project application needs to be considered.

    I have a concern that some AT users feed poor DG data into AT, and tell me that DG doesn’t work. In this case, they don’t have the productivity advantage of using DG only, plus, they make AT work harder causing more semi-manual work, which leads to productivity losses.

    I have a question for you:
    Would you agree that parallax free stereo viewing and image matching is possible with direct georeferencing with cameras above certain GSDs (and associated accuracy) and if the correct procedures are following? For example, projects with 15 cm GSD (and below) is when AT is definitely required. I would always use AT for engineering-grade projects.

    • Josef says:

      AT with direct georeferencing is sometimes leading to parallax free models and sometimes not. As one can not check thousand of models it is always good to do an AT before one spends hours for checking.

  4. Mathieu says:

    Hello, how do you handle/recommand to handle the deviation in vertical : the difference between normal to ellipsoid and normal to geoid ?
    As far as i know IMU make their measurements with respect to gravity (normal to geoid) and what you usually need in the end is the normal to ellipsoid for lidar reprojection at least. The difference can easily be greater than the angular accuracy of IMU’s or calibration stability. Modelling this difference is not so easy, the EGM2008 model still have rather large errors with reality.

    Regards,
    Mathieu

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: