I'm trying to convert hundreds of images that
Have an unknown subject centered in the image
Have a white background
I've used ImageMagick's convert utility in the following way
convert ORIGINAL.jpg -fuzz 2% -matte -transparent "#FFFFFF" TRANSPARENT.png
The problem is, some of my subjects are within the "white" scale, so, just like the weatherman wearing a green tie, some of my subjects seem to be disitegrating.
Is there any way to solve this via ImageMagick? Are there any alternative solutions? Scripting GIMP?
As you said, GIMP has a magic wand tool that can be used to select continuous areas of the same color, and so it can avoid the "green tie syndrome". The problem is that it may introduce a problem if there is something like a human hair crossing the image (that will seperate some of the white areas). Another common problem, especially with pictures of people, is when they put their hand next to the body and between the hand and the body there is a small hole.
Basically, it is not too hard to create a GIMP script that opens in batch many images, uses the magic wand to select the pixel at some corner (or if desired, in several known fixed places, not just one) and then removes the selection.
If it's hard to find a white area at a fixed spot, it is possible to do a search inside - meaning that the script searches for a white pixel on the borders, and it goes inside gradually in a spiral untill it finds some white pixel. But this is very very unefficient in the basic scripting engine, so I hope you don't need this.
If any of the suggested options above is OK, tell me and I'll create a gimp script for it. It will be even better if you can post some samples images, but I'll try to help even without these.
Related
I'm looking for a way to locate known text within an image.
Specifically, I'm trying to create a tool convert a set of scanned pages into PDFs that support searching and copy+paste. I understand how this is usually done: OCR the page, retaining the position of the text, and then add the text as an invisible layer to the PDF. Acrobat has this functionality built in, and tesseract can output hOCR files (containing the recognized text along with its location), which can be used by hocr2pdf to generate a text layer.
Unfortunately, my source images are rather low quality (at most 150 DPI, with plenty of JPEG artifacts, and non-solid backgrounds behind some of the text), leading to pretty poor OCR results. However, I do have the a copy of the text (sans pictures and layout) that appears on each page.
Matching already known text to it's location on a scanned page seems like it would be much easier to do accurately, but I failed to discover any software with this capability built-in. How can I leverage existing software to do this?
Edit: The text varies in size and font, though passages of it are consistent.
The thought that springs to mind for me would be a cross-correlation. So, I would take the list of words that you know occur on the page and render them one at a time onto a canvas to create a picture of that word. You would need to use a similar font and size as the words in the document - which is what I asked in my comment. Then I would run a normalised cross-correlation of the picture of the word against the scanned image to see where it occurs. I would do all that with ImageMagick which is available for Windows and OSX (use homebrew on OS X) and included in most Linux distros.
So, let's take a screengrab of the second paragraph of your question and look for the word pretty - where you mention pretty poor OCR.
First, you need to render the word pretty onto a white background. The command will be something like this:
convert -background white -fill black -font Times -pointsize 14 label:pretty word.png
Result:
Then perform a normalised cross-correlation using Fred Weinhaus's script from here like this:
normcrosscorr -p word.png scan.png correlation-result.png
Match Coords: (504,30) And Score In Range 0 to 1: (0.999803)
and you can see the coordinates of the match are 504,30.
Result:
Another Idea
Another idea might be to take Google's Tesseract-OCR and replace the standard dictionary with the text file containing the words on the page you are processing...
The grid is a blue/green. The work is in a black ink, and has a fair bit of variety of pressures, which I want to retain.
Here's a link to a small selection.
I have Photoshop v3
My attempts have involved using Select, Color Range, and sampling some grid, then inverting.
Is there a better way?
I also have some experience with Python and PIL, if that's a useful alternative.
This is a Photoshop answer, rather than a programming answer, but that seems to match your question's needs.
I applied a Black and White filter, and enabled a Blue filter, then set the Blue channel sensitivity to 300%, like this in Photoshop CC.
and got pretty good results like this:
In an older vsersion of Photoshop, you may need to go to Image->Mode->Lab Color and then go into the Channels palette and deselect Lab leaving just a and b channels selected, then use Select->Color Range to get the blues (or maybe the blacks!!!!) before going back to RGB mode.
I decided I wanted to change my gravatar to be circular. I have it circular on my blog with css and prefer the effect so decided to use a bit of imagemagick to give my image a circular alpha channel (so I could have it on SO, e.t.c. as well). a couple of quick commands later I was sorted:
# crop to square 422x422 with 0x0 offset
convert mike_gravatar.jpeg -crop 422x422+0+0 mike_gravatar_square.jpeg
# give circular alpha channel around it NOTE this is for a 422x422 image (note the 211x211)
convert mike_gravatar_square.jpeg \( +clone -threshold -1 -negate -fill white -draw "circle 211,211,211,0" \) -alpha off -compose copy_opacity -composite mike_gravatar_circle.png
brilliant, now we just upload this to gravatar, I will have a nice circular cropped image and all will be well with the world.
as you have probably guessed from the question all is not well:
ok, I must have clearly messed up my imagemagick and not checked that the before and after image are the same, reopen the images next to one another, and see that they are indeed the same. I try uploading again to gravatar and notice that they seem to process the images after the "cropping" stage, here is what it looks like in the browser after the file upload (before the cropping messes it up):
Alright, lets do some digging, someone else must have stumbled upon this before, so I have a look around and one lone soul in a desolate forum wasteland cries out. Now there is no response to this, but the relevant text is here:
It seems that if a photo or picture uploaded to Gravatar's cropper
doesn't have jet black, it will auto-level the nearest grey to black
and darken the whole image, including darkening whites into greys. Can
confirm that this occurs with any PNG image that has a grey background
or has a large enough proportion of it, whether or not it has 255
whites and regardless if it has alpha-blending or not
So it seems like I can fix this by putting in a single black pixel, that sounds alright so I try adding a black pixel, then a single black and a single white pixel, result:
So basically now I'm out of ideas:
does anyone have any idea what post-processing gravatar does, so I can undo it or counteract it's effects with pre-processing?
is this "feature" documented anywhere, or can it be turned off, or gotten around?
I think it would be quite cool to preprocess the image to counteract the darkening they would do to it but that would require knowing exactly what they do in order to change things and obviously might not be possible (depends on the relative movement of each colour, I suppose)
EDIT:
I tried making an inverse image to see if it was basing the processing on the average or the extreme values and that was also darkened, it seems that it's more likely to be the average:
Alright, I've got a solution that "worked for me" unfortunately it is just empirical and "good enough" (I'd quite like to check what's actually happening but unfortunately don't have any more time to devote to being nerd sniped) I'm gonna post what I did, what I think might have happened and how I think this should be solved if someone has enough time.
First off what I did, I simply tried whitening the image by random amounts, what ended up working was a gamma around 2
convert mike_gravatar_circle.png -gamma 2 mike_gravatar_circle_light_2.png
here is what the picture looks like both before and after processing by gravatar:
I feel it's pretty ridiculous that I need to clobber my picture like I do on the left to make it look normal so I'm going to leave this question open to see if anyone can show me a better/cleaner way of doing this.
EDIT: forgot to mention my (completely unfounded) guesses as to how this should be solved. So my guess is that gravatar might try and make the average color of the image some type of midrange value (as that might seem sensible... I guess, I don't know) and picks up the alpha as being all white. trying some experiments to determine could be interesting, but only if they had an api to automate uploading and downloading the images or it would be painful effort, I'm looking forward to any suggestions as to what people think is happening.
I want to replace the particular color of an image with other user selected color. While replacing color of image, I want to maintain the gradient effect of that original color. for example see the attached images.
I have tried to do so with CoreGraphics & I got success to replace color. But the replacing color do not maintain the gradient effect of the original color in the image.
Can someone help me on this? Is the CoreGraphics is right way to do this?
Thanks in advance.
After some struggling almost with the same problem (but with NSImage), made a category for replacing colors in NSImage which uses ColorCube CIFilter.
https://github.com/braginets/NSImage-replace-color
inspired by this code for UIImage (also uses CIColorCube):
https://github.com/vhbit/ColorCubeSample
I do a lot of color transfer/blend/replacement/swapping between images in my projects and have found the following publications very useful, both by Erik Reinhard:
Color Transfer Between Images
Real-Time Color Blending of Rendered and Captured Video
Unfortunately I can't post any source code (or images) right now because the results are being submitted to an upcoming conference, but I have implemented variations of the above algorithms with very pleasing results. I'm sure with some tweaks (and a bit of patience) you might be able to get what you're after!
EDIT:
Furthermore, the real challenge will lie in separating the different picture elements (e.g. isolating the wall). This is not unlike Photoshop's magic wand tool which obviously requires a lot of processing power and complex algorithms (and is still not perfect).
I'm working on a website design in gimp. I really love the figures and math equations at University of Iowa's stats website (http://www.stat.uiowa.edu/). I would like to create something like this, perhaps do handrawn equations and figures and scan it in.
Once I have the appropriate image, how could I turn it into a transparent image that I can overlay onto a header (in gimp)?
Thanks
I would scan some diagrams and handwritten math equations . Then I would open it from gimp. After you done it you can go to --> colours and use Desaturate. You can use any method from there. There are only 3. See which one looks the best on your photo. After you got black and white photo go to colours and choose --> Posterise and use it with level set to 2. After this you go to --> Layers --> Transparency and choose Colours to Alpha. Choose white to remove the background. After this save you image as . png file. You should scan possible black and white image. You can write some diagrams on white paper without lines of print some diagrams and math equation. If you got scanner you can also set the settings to scann only in two colours and then you do not have to use Desaturation. I hope this will help you.