Hi GuysI have a graph image (JPEG) which I'm reading into a matrix of rgb components. However the graph points (in red) comes in various shades of red pixels around the real value. (Example - in one column, instead of 1 red pixel, I see around 20 pixels of different shades of red). Could anyone point me to some filtering methods that I can use to group these pixels in either red or white (maybe fourier method?) If I can transform the image to 1shade of black, white and red, it would be easy to pick the data points from the image.I'm doing this in R using readJPEG function. I dont mind implementing the method if there is no library that already does it.Thx

If you're doing image analysis the Y component of the RGB -> YUV conversion might be of more use. In fact, JPEGs are natively YUV.Anyway, your problem is that the jpeg compression has introduced artefacts into the image? Is the real value RGB(255, 0, 0) or has the compression changed that too?

Not 100% sure I understand your question, but if you're looking to back out the the set of x,y coordinates corresponding to the points of a plot for which you have only a jpeg then you may want to look at the Google Chrome plugin named WebPlotDigitizer.

Isn't there possible solutions with Mathematica?Link1Link2Alan?