[GRASS-dev] Color based image segmentation + Text layer detection + Extract text as a vector layer

Dear All,

I would like to propose my professor and I research result as a very useful component for Grass. I want to divide my proposal in three parts. First, implementing color based image segmentation using k-means algorithm and then without user interaction try automatically detect the best number of layers in the map (automatically detected the best cluster number). Second, between all extracted layers find the text layer and do some image processing filters on that layer to make it more readable. Third, I want use the detected text layer and pass it to the open source OCR software, Tesseract, and save the result as a vector layer.
I really appreciate any comments and suggestions.

Regards,
Ramtin Boustani