PureBasic Forums - English

dear all,
i need to code a procedure like image entropy calculator, for more clear please take a look at this:

https://easyupload.io/vstl9o

In this file i include 6 image files, 1 good and 5 wrong, all are not complete image, only a crop, for example purphose, i need to identify with the major possible percentage the good and the fail image, for now i've tried to code a procedure for count color image, but i notice that some good image have a lot of color like the wrong one, after i try to check alternage of color on single line, but also this return me fake results, so i think that an image entropy calculator can be a good solution, but i don't found some clear algorithm to better know about this.

Anyone have a problem like me and already found a possible solution ?

Please ignore color in image that of course can change on all possible colors in both good and wrong case.

Thanks to all

Maybe is better if you use OpenCV for this job

Caronte3D wrote: Fri Feb 24, 2023 12:38 pm Maybe is better if you use OpenCV for this job

i know, but i prefer to not use external elements in this project

If your resized ( Or zoomed) the image the bad image will be pixalized heavily
So you can compare between the no of colors of a segment area of the main image and the equal size of the resized image it may give you an opinion about the quality
Have a good luck

RASHAD wrote: Fri Feb 24, 2023 1:45 pm If your resized ( Or zoomed) the image the bad image will be pixalized heavily
So you can compare between the no of colors of a segment area of the main image and the equal size of the resized image it may give you an opinion about the quality
Have a good luck

interesting idea, i'll take a look on next week about this, thanks!

the problem is not so easy to solve, because i notice that also good image can have huge entropy, still thinking about a possible workaround, thanks to all