How to Georeference a Floorplan: A Practical Guide

Georeferencing is the process of taking an image that has no inherent connection to a real-world location — a scanned floorplan, a PDF site plan, an old blueprint — and assigning it real coordinates, so it lines up correctly with maps, satellite imagery, and other spatial data.

What georeferencing actually means

A floorplan file, on its own, is just pixels. It has a width and a height in pixels, but nothing telling a computer where on Earth it belongs. Georeferencing fixes that by mathematically mapping pixel coordinates in the image to longitude/latitude (or a projected coordinate system) in the real world — so a GIS tool, a map viewer, or a piece of survey software can place it accurately.

Why it matters

Once a floorplan is georeferenced, it stops being a static image and becomes usable spatial data. That unlocks a handful of common workflows:

• Facilities and campus GIS teams overlaying building interiors on a broader site map
• Emergency response and disaster planning, where responders need interior layouts positioned accurately relative to the outside world
• Permitting and code compliance, where a plan needs to be checked against zoning or parcel boundaries
• As-built documentation, comparing a designed layout against what's actually on the ground
• Any workflow that needs to combine a floorplan with satellite imagery, parcel data, or other GIS layers

The basic workflow

1. Start with a clean image of the floorplan — a PDF, JPG, or PNG, ideally with minimal clutter around the actual building outline
2. Identify a handful of recognizable reference points on the plan — corners, doorways, distinctive features you can also spot in satellite imagery
3. Match each of those points to its real-world coordinate by clicking the same feature on a satellite map
4. Let the software compute a transform — the mathematical relationship between pixel space and real-world coordinates
5. Export the result in whatever format your downstream tools need

How many reference points do you actually need

This is the part most guides skip, and it matters more than people expect:

• 1 point — only gives you position, not scale or rotation. Rarely enough on its own.
• 2 points — enough for an exact-fit similarity transform: position, uniform scale, and rotation, with both points locked exactly in place.
• 3 points — enough for a full affine transform, which additionally allows independent scaling on each axis. All three points fit exactly.
• 4+ points — a least-squares best fit. More points generally means more accuracy, but no single global transform can pass through more than 3 arbitrary points exactly, so with 4 or more the fit gets as close as mathematically possible rather than locking every point perfectly.

In practice, 3 well-chosen points scattered across the plan (not clustered in one corner) usually gives a solid, trustworthy result.

Common mistakes

• Picking reference points that are ambiguous in satellite imagery (a generic corner with no distinguishing feature nearby)
• Clustering all your points close together, which amplifies any small error across the rest of the plan
• Ignoring rotation — buildings are rarely perfectly aligned to true north, and a 1- or 2-point setup can silently get this wrong
• Trusting the result blindly for anything with real legal or safety consequences without an independent check

Doing this without a full GIS suite

Traditional GIS software can do all of this, but it's often more setup than a one-off floorplan job actually needs. FloorGeo is built specifically for this narrower workflow — drop in a PDF or image, click a few reference points, and export directly to GeoTIFF, GeoJSON, Shapefile, or a plain CSV of coordinates, without needing to learn a full desktop GIS application first.