Tag Archives: Spectral estimation for fun and profit

Texture MTF

Based on Dead-leaves type of test targets* (updated)

All Matlab code covered by BSD License

Refined Measurement of Digital Image Texture Loss

Article: Refined Measurement of Digital Image Texture Loss

21 March 2024 Update: Based on suggestions from Dinesh Iyer at MathWorks, the code has now been updated to allow two methods,

  1. Two-Scan: using both an input reference (Scan) image of the test target and (output) test image. The method used in the article, and originally posted here.
  2. Model-Spectrum: based on an ideal computed spectrum for the deadleaves pattern (new).

Matlab source code to compute the measure of texture loss based on images of dead-leaves type of test targets. The method includes correction for image noise and shading/non-uniformity, as described in,
P. D. Burns, Refined Measurement of Digital Image Texture Loss, Proc. SPIE vol. 8293, Image Quality and System Performance IX, 2013,86530H, 2013

The method was also used in the evaluation of JPEG2000 image  compression in,
P. D. Burns and D. Williams, Measurement of Texture Loss for JPEG 2000 Compression, Proc. SPIE vol. 8293, Image Quality and System Performance IX, 2012

Software: 2024 Matlab function texture_MTF.m and supporting functions available here

Video demo.

Video demo.

 

Video: Demonstrating the running of the function here.
Download the 8 MB file if it does not run in your browser.

Refinements: I am distributing this code primarily for those who are familiar to adapting Matlab for their own applications. Depending on interest expressed, I can include more flexible parameter selection, data checking and other refinements that I would normally include as part of a complete image-based tool. Also, please contact me if you are interested in an executable version of this method as part of a console-type programme. I have fun with this sort of thing too.
_______________________
*Spectral estimation for fun and profit