From raster/pixel annotation to vector

[nmaisonneuve]

image of a paper with a line drawn by hand =>georeferenced polyline

[pamoakoy]

Can you explain further, what the issue is, Nico?

[nmaisonneuve]

This is part of my project on an augmented paper-based mapping: You take a picture with your phone of a paper annotated by people with their pens to extract, georeference and send the annotations to the GIS. Process People draw on a paper some annotations using different forms e.g. lines, circle , cross etc.and different colors. This paper could be originally a printed map on which people annotate new stuff to update the map or a white paper. In all the cases , the paper has an iD and is initially georeferenced (we know the coordinates of area represented). So 1) I take a picture , 2) I apply some technics to correct the perspective of the photo, 3) I extract only the annotations (easy when the paper is clean/white , a bit harder but ok when we have a printed map). At this point I have an image with only the annotations + the coordinates on the area represented by the corners of the paper. 4) Then I transform the pixel into vectors using recognition technics, 5) I according to the color or shaped detected I add tags to describe the semantic of the annotation(e.g. destroyed building, new road, hospital) , 6) send I send this information to the GIS server.

Current state 1 to 5 is done but 4 to 5 is not well done (I recognised only circle or cross). The goal is to extend this in a more general way. especially extracting the pixelised lines into a polyline structure with geocoordinates.

[nmaisonneuve]

oups I closed the issue

[pamoakoy]

Is the target here for "ordinary" volunteers or expert volunteers?

[nmaisonneuve]

ordinary (it's a side project so this request is not a real priority)

[nmaisonneuve]

ordinary (it's a side project so this request is not a real priority)