I'm trying to get rid of noise in the grayscale image. I tried the non-local-mean and Gaussian filter, but none of them give me good results. Is there any recommended algorithm I can use?
In Imagemagick, you can use the kuwahara non-linear edge preserving filter with different values (see https://en.wikipedia.org/wiki/Kuwahara_filter). Choose whichever looks best to you.
Input:
convert img.png -kuwahara 5 k5.png
convert img.png -kuwahara 9 k9.png
convert img.png -kuwahara 15 k15.png
convert img.png -kuwahara 25 k25.png
Not sure what you are hoping for, but here each pixel is replaced by the mean of the surrounding 49x49 pixels just using ImageMagick in the Terminal:
convert noise.png -statistic mean 49x49 result.png
Here's the median of the surrounding 25x25 pixels:
convert noise.png -statistic median 25x25 result.png
Related
Using Gimp, given an input image, I can improve its contrast using Colors > Hue Chroma... by setting Chroma=50 (in a scale between -100 and 100) and leaving Hue=0 and Lightness=0. So it appears I'm doing an HCL transformation.
Is there an equivalent command for Magick?
The following image shows the GIMP effect:
Image
Updated Answer
Not sure about this at all. I think you can get pretty close with -modulate if you go into an LCH colourspace, but I have no idea if it will work consistently. I got:
magick cXDv3.jpg -define modulate:colorspace=LCH -modulate 100,150 result.jpg
If that doesn't work, or is not to your liking, read on...
Generic Method for any GIMP filters
The method below should allow you to replicate any GIMP filter with ImageMagick - as long as it is a pure "point process", I mean one where each pixel's output value is purely derived from its input value and not an "area process" where surrounding pixels contribute - such as blurring or median filtering, for example.
It's called a HALD-CLUT. You would create a HALD-CLUT something like this:
magick hald:16 clut.png
Then take that file (clut.png) into GIMP and apply your GIMP processing on it and save the result as GIMP-H0-C50-L0.png so we know how GIMP affects each colour. You do that just once.
Then you go back to ImageMagick and apply that CLUT to your image:
magick input.png GIMP-H0-C50-L0.png -hald-clut result.png
That gives me this:
and I think you'll agree the left side looks pretty similar to the right side of your input image.
Original Answer
I don't know what that command does in GIMP, but you can convert to HCL colourspace in ImageMagick and select the Chroma channel for modification like this:
magick INPUT.PNG -colorspace HCL -channel G ...
You then want to do something ? to affect the Chroma channel, so try -auto-level for now, and then return to sRGB colourspace and save:
magick INPUT.PNG -colorspace HCL -channel G -auto-level +channel -colorspace sRGB RESULT.PNG
Then you need to provide more clues or experiment more with what that command does in GIMP - or provide examples.
I need to generate low resolution black and white images of texts in ImageMagick. These images will be plotted in a small LED matrix. The text need to have 7 pixels of height.
For now, I'm using:
convert -negate -threshold 15% -font Courier -size 80x11 caption:'hello' out.bmp
Output image:
Even with the height being more than I need, due to low resolution and anti-aliasing correction, the letters are not pretty and symmetric. Has anyone did this and can help me out?
Version: ImageMagick 6.8.9-9 Q16 x86_64
The solution I found was to use an TrueType font. Just got a free font from the internet and used it in the size it was built for.
P.S.: Switched for OpenCV as well. My Python app generates images dynamically. The cost for invoking ImageMagick several times(could get close to hundred) per minute is too high.
Posting a snippet, hope it helps.
import cv2 as cv
from PIL import ImageFont, ImageDraw, Image
# Creates a black image with 80x10 size
img = Image.new('RGBA', (80, 10), (0,0,0,0))
draw = ImageDraw.Draw(img)
# Load TrueType font of height size 8
font = ImageFont.truetype("font.ttf", 8)
# Draw text using the loaded font
draw.text((0, 0), "Hello World!", font=font)
img.save("out.bmp")
Output Image:
I would be inclined to output the letters larger than required, then to trim any extraneous spare space so as to make the most of the available resolution, then resize down to your specific needs:
convert -size 320x32 -font Courier label:'hello' -trim +repage -resize 80x8 +write out.gif
Mark, I think he wants a binary result. But you have an excellent idea.
Let's take Mark's result, threshold and then scaling down to 8 pixels tall. This ImageMagick command seems to work better than my earlier post.
Mark's Output:
convert wcwuj.gif -threshold 60% +write thresh.gif -scale x8 result.gif
Threshold Result:
Scaled Result:
Perhaps making Mark's image much larger and choosing a better threshold will produce a better result.
You have not told us what version of ImageMagick nor platform and you do not show your result for us to see what might be wrong. Also your ImageMagick syntax is not proper, though ImageMagick 6 is rather forgiving.
This is what I get using ImageMagick 6.9.10.8 Q16 Mac OSX Sierra. The first output is 8 pixels tall and the second output is scaled by 1000% (10x).
This forum does not seem to convert bmp to a usable format for display, so I am using GIF in place of BMP. But my results look the same whether BMP or GIF
convert -size x8 -font Courier label:'hello' -negate -threshold 20% +write out.gif -scale 1000% out2.gif
I have tried changing threshold, but much larger or smaller values make it worse. A range from about 10-30% produces the same results.
I have also tried using -monochrome in place of -threshold and get the following:
convert -size x8 -font Courier label:'hello' -negate -monochrome +write out3.gif -scale 1000% out4.gif
You might try a dot-matrix type font. See https://www.1001fonts.com/digital+dot-matrix-fonts.html?page=1&items=10. I have not tried any of them.
You could try some of the old X11 fonts. These were hand-drawn rather than being rendered from a set of curves, so they look good at very small sizes.
For example, if I run xfontsel I get things like this (enlarged for clarity):
Take a look in /usr/share/fonts/X11/misc.
I was able to convert my EXR image to a PNG using the techniques outlined in Image conversion from IFF and EXR formats to JPEG format .
convert 0007.exr /tmp/0007.png
Unfortunately the PNG looks quite dim.
What should I add to the imagemagick convert command line to increase the brightness?
Starting with this:
You could try -auto-gamma:
convert start.jpg -auto-gamma result.jpg
If the -auto-gamma overcooks the image for your liking, you could apply a percentage of it. So, here I clone the original image and apply auto-gamma to the clone but then only blend 80% back into the original because I feel auto-gamma overdoes it:
convert start.jpg \( +clone -auto-gamma \) \
-define compose:args=80 -compose blend -composite result.jpg
Or, another option, you could experiment with your particular images and maybe try using -modulate for the brightness, where 100% means "do nothing", so numbers over 100 increase the brightness:
convert start.jpg -define modulate:colorspace=LCHuv -modulate 160 result.jpg
You can try -auto-level, which will take the minimal value and the maximal value of your picture and then stretches the values to the full range of values:
convert input.exr -auto-level output.jpg
Note that if you picture was too bright and this does not help, then it might be that your image is stored with 32 Bit, while ImageMagick is working with 16 Bit and no HDRI support. 32 Bit input is supported if convert --version
either show Q32 as part of the version string or lists HDRI under features.
Depending on your operating system you might be able to install another variant of ImageMagick. For example, for Debian Buster we can use sudo apt list imagemagick* to see that the package imagemagick-6.q16hdri is available. Installing this package provides convert-im6.q16hdri, which allows reading 32 Bit EXR images.
EXR is in linear RGB colorspace. You want to convert it to non-linear sRGB colorspace in Imagemagick as:
Input:
convert image.exr -set colorspace RGB -colorspace sRGB output.png
I known some ways to compare 2 images with ImageMagick or OpenCV
Using Objective-C, is there any way to compare two images and get a % difference value returned?
http://www.imagemagick.org/Usage/compare/#methods
How can I quantify difference between two images?
Image comparison - fast algorithm.
But for my case, I also have the same character with different position.
Image1:
Image2:
or
or
So, what should I do now to find the % difference value between Image1 and Image2?
This actually answers your question - which doesn't in fact ask anything about images 3 and 4 - but I fear it will not help you much.
As #GPPK suggests, you need to trim the extraneous material off around your kanji characters, which you can do with the -trim command in ImageMagick. I have added a thin red border so you can see where the edges are:
convert kanji2.png -trim kanji2-t.png
If you want do that to images 1 and 2, and then compare them, you can do that all in one go like this:
convert -metric ae kanji1.png kanji2.png -trim -compare -format "%[distortion]" info:
0
which shows there are zero pixels different in the resulting images if you trim kanji1 and kanji2.
If you compare the trimmed kanji1 and kanji3 like this, you get:
convert -metric AE kanji1.png kanji3.png -trim -compare -format "%[distortion]" info:
893184
which indicates 900,000 pixels of 5,000,000 are different.
Likewise, if you compare kanji1 and kanji4:
convert -metric AE kanji1.png kanji4.png -trim -compare -format "%[distortion]" info:
1.14526e+06
or 1.1 million of 5 million.
But this doesn't help when your images are a different size (scale), or rotated.
You could scale your images to a normalised size before comparing, and I guess that might help you become a bit more "scale invariant":
convert -metric AE kanji1.png kanji4.png -trim -scale 1000x1000! -compare -format "%[distortion]" info:
You could also rotate your images using a little iterative procedure that rotates the images through say +/- 20 degrees and chooses the one with the smallest trimmed bounding box to become a little more "orientation invariant". But then you will still have a problem if the characters are sheared, or fatter, or thinner, or brighter, or darker, or contrastier... I think you need to look into "Template Matching".
I have image like this from my windstation
I have tried get thoose lines recognized, but lost becuase all filters not recognize lines.
Any ideas what i have use to get it black&white with at least some needed lines?
Typical detection result is something like this:
I need detect edges of digit, which seams not recognized with almost any settings.
This doesn't provide you with a complete guide as to how to solve your image processing question with opencv but it contains some hints and observations that may help you get there. My weapon of choice is ImageMagick, which is installed on most Linux distros and is available for OS X and Windows.
Firstly, I note you have date and time across the top and you haven't cropped correctly at the lower right hand side - these extraneous pixels will affect contrast stretches, so I crop them off.
Secondly, I separate your image in 3 channels - R, G and B and look at them all. The R and B channels are very noisy, so I would probably go with the Green channel. Alternatively, the Lightness channel is pretty reasonable if you go to HSL mode and discard the Hue and Saturation.
convert display.jpg -separate channel.jpg
Red
Green
Blue
Now make a histogram to look at the tonal distribution:
convert display.jpg -crop 500x300+0+80 -colorspace hsl -separate -delete 0,1 -format %c histogram:png:ahistogram.png
Now I can see all your data are down the dark, left-hand end of the histogram, so I do a contrast stretch and a median filter to remove the noise
convert display.jpg -crop 500x300+0+80 -colorspace hsl -separate -delete 0,1 -median 9x9 -normalize -level 0%,40% z.jpg
And a final threshold to get black and white...
convert display.jpg -crop 500x300+0+80 -colorspace hsl -separate -delete 0,1 -median 9x9 -normalize -level 0%,40% -threshold 60% z.jpg
Of course, you can diddle around with the numbers and levels, but there may be a couple of ideas in there that you can develop... in OpenCV or ImageMagick.