Legacy ImageMagick Discussions Archive

If you can redirect the output of ghostscript to a pipe, you can then handle it with "GHOSTSCRIPT_COMMAND | convert - -resize 25% outfile.png" . use "-" instead of the input filename After that, I run mogrify three times (and if there is a way around this I'd be VERY happy) to cr...

Just curious about your python code for creating the hough transform. I don't read code that well, but it looks like you need openCV to create the accumulator? or am I mistaken? can that be done simply in a python or shell script without openCV? For the accumulator itself I do not need openCv per s...

Had not considered that. I don't understand why it is quantized to 8 directions only if you are using arctan2(DY,DX) for getting theta. Should you not get the actual angle of the lines? I need to think on this more. with binary data the 3x3 sobelX and the sobelY can have any integer value from -4 t...

I would not necessarily expect it to work as you are doing and edge extraction on an edge image. Perhaps better to try from the binary image I made before extracting the edge. Here it is: Not much better actually. As far as I can tell the problem is mainly the binary nature of the image. With a 3x3...

why 5 (not 4) bright spots? Is that due to the open pages on the left side of the book? Not sure actually, it might be the middle of the book, or one of the other lines. I still have to finish the code that finds the local maxima in hough space and converts them back into lines. I did not do any no...

Some more experiments. I went back to python since I do not see a way to make the previous attempt work at acceptable speed and quality. My conclusion is that the oriented hough transform is indeed much better suited than the classical for this particular problem. Still not sure if this could or sho...

Well, the idea was this: - Generate an image with the theta values of all edge pixels (the orientation of the gradient) - Generate an image with the rho values of all edge pixels (The shortest distance to the origin of a line throught that pixel with orientation theta) - create a 2d histogram: count...

I am not sure I understand what you want. but converting all bmp files to png in the current directory on windows is done with FOR %a in (*.bmp) DO convert %a %~na.png If you put this line in a batch file, you have to double all %: FOR %%a in (*.bmp) DO convert %%a %%~na.png deleting can be done aft...

I have been thinking about doing a hough transform with the tools already provided by imagemagick, and it seems relatively straightforward. All that is missing is a way to simulate the accumulator array. This is just a random idea, but is there some way to do a 2d histogram in imagemagick? I have se...

that is correct output because you are really doing three compose operations ( background + logo: ) + logo: ) -> white I see, that makes sense, thanks! I assumed the background was always an implicit "under" composition, no matter what. Do i have to set the background depending on what ty...

"-compose plus -layers flatten" alway produces a blank white image Example: convert logo: logo: -compose plus -layers flatten test.png http://img20.imageshack.us/img20/401/testyxg.th.png Blank image convert logo: logo: -compose plus -composite test.png http://img20.imageshack.us/img20/718/...

But the fact that it uses openCV and perhaps Python is going to limit its use with IM users. But don't let this comment deter you. Use what you need to get the job done. Well the idea was that when this program is released, the imagemagick and openCV libraries can be bundled with it, so the user on...

The colored lines are the debugging output from my script. It is about 400 lines, could be shorter, and uses openCV for canny edge detect, hough transform and graphic display. It uses imagemagick, but currently by calling sobelseries.bat and convert.exe through the shell. I want to use the imagemagi...

Some tests with whiteboards. cleaning was done with for %a in (*-hough.*) do convert "%a" ^ ( +clone -scale 12.5% -blur 0x30 -resize 800% ) ^ +swap -compose divide -composite ^ ( -clone 0 -contrast-stretch 2%,50% -threshold 60% ^ -blur 4x2 -threshold 99% ^ -blur 4x2 ^ ) -compose screen -co...

I have not had time to get into those equations yet, but understand the principles. So how do you plan to avoid the divide by zero or get the focal length if you don't have that? Is that for any two sides or the two sides that are involved with perspective? If the latter, then if you find parallel,...