Does a Quad Bayer sensor change the sensor size used for georeferencing calculations?
Asked 5/12/2023
4 views
2 answers
0
I'm georeferencing DJI drone images from EXIF data and need the camera sensor size, focal length, and image dimensions to calculate image footprints. This works on several DJI models, but with the Mini 3 Pro my calculated footprints are much too large.
I found the Mini 3 Pro sensor listed as about 10 x 7.5 mm. The camera can produce 48 MP images, but my test images are 12 MP JPEGs. Because this is a Quad Bayer sensor, I wondered whether I should change the effective sensor size in my calculations.
I tried using a smaller sensor size and even dividing the listed sensor dimensions by 4, but that made the results wrong in the other direction.
For footprint/georeferencing math, should a Quad Bayer sensor change the sensor dimensions at all, or is the only adjustment that each sensor dimension should be treated as half-resolution rather than quarter-resolution when converting between the 48 MP array and 12 MP output?
Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0
Photography Stack Exchange contributor
3y ago
2 Answers
2
There is small mistake with the size of the Bayer matrix. It is 2x2 cells so you should divide each dimension by 2, not by 4.
P.S. for information Fujifilm X-Trans matrix is different, it is 3x3 so if you have/use such sensor you should change the calculations.
Originally by user34947. Source · Licensed CC BY-SA 4.0
user34947
3y ago
0
Generated from our catalog & community — verify before relying on it.
A Quad Bayer sensor does not mean the physical sensor is one-quarter the size. The array is grouped in 2×2 cells, so if you need to relate the full pixel grid to the 12 MP output, each dimension is effectively halved, not divided by 4.
So:
- don’t divide the sensor width and height by 4
- if you are adjusting for Quad Bayer binning, think in terms of 2×2 grouping
- the physical sensor size itself remains the same
In other words, Quad Bayer changes how pixels are read/combined, not the actual CMOS dimensions. If your code assumes the 12 MP image comes from a physically smaller sensor, the footprint will be wrong.
The community answer indicates the specific mistake is using quarter-size dimensions instead of half-size dimensions for each axis. A Quad Bayer pattern is 2×2, so divide each dimension by 2 if your model needs an effective per-axis sampling adjustment.
Also note that other color filter layouts can differ; for example, Fuji X-Trans uses a different repeating pattern.
Recommended products
UniqueBot
AI3y ago
Your Answer
Related Questions
Why does the Sony CX405 list different still-image megapixel counts in its specifications?
Does circle of confusion depend on megapixels or sensor size?
Which pixel count should you use when comparing camera resolution?
What does Snapsort's "true resolution" mean compared with megapixels?
How can sensor dimensions be estimated from pixel size, megapixels, and aspect ratio?