Scan PDF exported non spatial -Image file - Georeerence and extract the points with the size of the colour available

​@takashi 

​@kailinatsafe 

and experts 

need support n suggestions of expertise.

I shall be thankful to u 

This is not something I would fix using FME. For georeferencing my go to tool is QGIS, it has a great georeferencing tool.

​@nielsgerrits 

thank you for the response.

I have like many images in bulk so that the I was trying to find a solution through FME .

If it was like few then I would have opted the ArcGIS Pro or QGIS 

Hi ​@abnan0001 

 

I used the transformer RasterGeoreferencer in your source JPG:

 

Thanks in Advance,

Danilo

Thank you 

Source to georeference them using extent layer because. Location is actually Saudi Arabia .

Need your kind guidance to resolve it.

thank you once again 

 

As far as I know there is no easy solution for stuff like this.

How many prints are we talking about? 10? 100? 1000? 10.000? 100.000? To get an idea how much effort it would be to do it manually and when automating things will pay off.

Are all scans exactly the same? Can we find the transformation for one print and use it for all prints?

If not, is the layout the same? The quality, the grid, the degrees on top and on the right.

If so, you might be able to isolate the grid, OCR the numbers, match the numbers to the grid and go from there. It is not impossible, but it will take some work.

​@nielsgerrits and ​@danilo_fme 

Thank you for your kind support

Let me elaborate in more detail

There are about 1K scanned pdfs that we are converting to img/ tiff and then georeference with same extent and having same grid for all.

The  Goal is to georeference and then extract the lat/lon values for the color points based on their size like available in the attach map in blue and red in the provided link.

thank you

Then do one print manually, find the coordinates of the corners and then use the RasterGeoreferencer as ​@danilo_fme suggests to get all prints on coordinates.

Ok cool understood for georeference  and then how can get the lat n kin value with  the size and color so that I can differentiate then.

any suggestions on this 

I’m not sure if this is the best way to do it, but what I would try is to use the RasterExpressionEvaluator (which is a b!tch to configure) to set all pixels which are not data to nodata. Then use the RasterCellCoercer with ignore nulls to get the areas from the pixels and go from there.