Meaningful Scales Detection: an Unsupervised Noise Detection Algorithm for Digital Contours
Please cite the reference article if you publish results obtained with this online demo.

This demonstration applies the meaningful scale detection on discrete contours extracted from grayscale image.

Select Data

Click on an image to use it as the algorithm input.

polygon with some noise
flower with some noise
star with some noise
image S
image T
ellipse noise

image credits

Upload Data

Upload your own image files to use as the algorithm input.


Images larger than 16777216 pixels will be resized. Upload size is limited to 16MB per image file and 10MB for the whole upload set .
TIFF, JPEG, PNG, GIF, PNM (and other standard formats) are supported. The uploaded will be publicly archived unless you switch to private mode on the result page.
Only upload suitable images. See the copyright and legal conditions for details.