MATLAB Answers

Ann G

How can I find the pixel size of an image?

Asked by Ann G
on 5 Apr 2016
Latest activity Commented on by Walter Roberson
on 17 Aug 2017
I want to know the pixel size to calculate the size of an image...
number of pixels * pixel size


Sign in to comment.

2 Answers

Answer by Image Analyst
on 5 Apr 2016
 Accepted Answer

See my spatial calibration demo. If you know the size of something in the image, such as a ruler, then you can spatially calibrate your image. See attached demo.


Somehow I got this message:
Undefined function or variable 'Calibrate'.
function Calibrate is part of that .m file, so it should be found if you copied in the entire .m file.

Sign in to comment.

Answer by Walter Roberson
on 5 Apr 2016

In general you cannot do that, and a lot of the time when you can get some information, the information is just plain wrong. But the times you can get information, it would be stored in either EXIF headers (JPEG, TIFF) or TIFF headers (TIFF). The EXIF information, when present, can be retrieved by using imfinfo() and looking at the DigitalCamera field of the returned structure (note: the field might not be present.) TIFF headers that are not automatically synthesized to EXIF need to be read using the Tiff class.


Thank u!!
Note that Image Analyst has a demo on Spatial Calibration -- that is, if you can point to two things in the image and say how far apart they are "really", then his demo can calculate any other distances on the image. This does not rely on any information stored in the file, but it does rely upon there being knowledge of "real" sizes of something in the picture.
Also, when you are using a camera that is designed as a scientific instrument, or is sufficiently high-end (probably not consumer level), then there might be enough EXIF or TIFF headers stored to be able to calculate some distances without knowing the "real" size of something in the image. That calculation requires information about the aperture size and lens focal length, and some depth-of-field calculations, and might require some information about how far away the autofocus thought the object was.

Sign in to comment.