Hey, i'm using a program called pngquant to convert 24 bit PNGs to 8-bit PNGs. Everything seems to work fine, and I don't notice any loss of quality for icons and other images that don't contain too much colors. Now when I feed it a PNG photo with zillions of colours, it produces a PNG8 where I can see some quality loss.
I'd like to determine that quality loss programmatically. I'd like to know when converting a PNG24 to PNG8 is safe or not. Sort of what webpagetest.org does -- they tell you that this specific image will be smaller in size if converted to PNG8 and will not loose quality.
Any ideas?
Thanks.
This sounds like a full-reference image quality assessment problem.
The simplest way to approach this is to try computing the PSNR between the PNG24 and PNG8 images. This is a measure of the difference between the two images. The higher the PSNR, the less different the images are. After using your color quantization software, check if the PSNR is above some threshold (you'll have to determine that empirically), and if it is, then the quantization was "safe".
PSNR has its down sides, namely the fact that it doesn't always correspond to the way the human visual system works (for example, it neglects the phenomenon of spatial and contrast masking). Another metric, SSIM, attempts to take care of that problem, but is slightly more difficult to compute (here is an OpenCV implementation, though). You can use SSIM instead of PSNR in the thresholding approach I described above.
Here's another thread which you might find useful.
Quite simple. If the image you are converting from PNG24 to PNG8 has more thant 256 colors, you gonna loose quality. Do I missed something?
For development of pngquant I use my own SSIM tool, since the OpenCV-based one didn't seem to support gamma correction nor alpha channel properly.
When you run pngquant -v it will output amount of error introduced as MSE=n (n is mean square error — 0 is perfect quality).
The latest version has --quality setting which lets you set minimum required quality. If it can't achieve it, it won't save the file.
Related
I have a bunch of poor quality photos that I extracted from a pdf. Somebody I know has the good quality photo's somewhere on her computer(Mac), but it's my understanding that it will be difficult to find them.
I would like to
loop through each poor quality photo
perform a reverse image search using each poor quality photo as the query image and using this persons computer as the database to search for the higher quality images
and create a copy of each high quality image in one destination folder.
Example pseudocode
for each image in poorQualityImages:
search ./macComputer for a higherQualityImage of image
copy higherQualityImage to ./higherQualityImages
I need to perform this action once.
I am looking for a tool, github repo or library which can perform this functionality more so than a deep understanding of content based image retrieval.
There's a post on reddit where someone was trying to do something similar
imgdupes is a program which seems like it almost achieves this, but I do not want to delete the duplicates, I want to copy the highest quality duplicate to a destination folder
Update
Emailed my previous image processing prof and he sent me this
Off the top of my head, nothing out of the box.
No guaranteed solution here, but you can narrow the search space.
You’d need a little program that outputs the MSE or SSIM similarity
index between two images, and then write another program or shell
script that scans the hard drive and computes the MSE between each
image on the hard drive and each query image, then check the images
with the top X percent similarity score.
Something like that. Still not maybe guaranteed to find everything
you want. And if the low quality images are of different pixel
dimensions than the high quality images, you’d have to do some image
scaling to get the similarity index. If the poor quality images have
different aspect ratios, that’s even worse.
So I think it’s not hard but not trivial either. The degree of
difficulty is partly dependent on the nature of the corruption in the
low quality images.
UPDATE
Github project I wrote which achieves what I want
What you are looking for is called image hashing
. In this answer you will find a basic explanation of the concept, as well as a go-to github repo for plug-and-play application.
Basic concept of Hashing
From the repo page: "We have developed a new image hash based on the Marr wavelet that computes a perceptual hash based on edge information with particular emphasis on corners. It has been shown that the human visual system makes special use of certain retinal cells to distinguish corner-like stimuli. It is the belief that this corner information can be used to distinguish digital images that motivates this approach. Basically, the edge information attained from the wavelet is compressed into a fixed length hash of 72 bytes. Binary quantization allows for relatively fast hamming distance computation between hashes. The following scatter plot shows the results on our standard corpus of images. The first plot shows the distances between each image and its attacked counterpart (e.g. the intra distances). The second plot shows the inter distances between altogether different images. While the hash is not designed to handle rotated images, notice how slight rotations still generally fall within a threshold range and thus can usually be matched as identical. However, the real advantage of this hash is for use with our mvp tree indexing structure. Since it is more descriptive than the dct hash (being 72 bytes in length vs. 8 bytes for the dct hash), there are much fewer false matches retrieved for image queries.
"
Another blogpost for an in-depth read, with an application example.
Available Code and Usage
A github repo can be found here. There are obviously more to be found.
After importing the package you can use it to generate and compare hashes:
>>> from PIL import Image
>>> import imagehash
>>> hash = imagehash.average_hash(Image.open('test.png'))
>>> print(hash)
d879f8f89b1bbf
>>> otherhash = imagehash.average_hash(Image.open('other.bmp'))
>>> print(otherhash)
ffff3720200ffff
>>> print(hash == otherhash)
False
>>> print(hash - otherhash)
36
The demo script find_similar_images also on the mentioned github, illustrates how to find similar images in a directory.
Premise
I'll focus my answer on the image processing part, as I believe implementation details e.g. traversing a file system is not the core of your problem. Also, all that follows is just my humble opinion, I am sure that there are better ways to retrieve your image of which I am not aware. Anyway, I agree with what your prof said and I'll follow the same line of thought, so I'll share some ideas on possible similarity indexes you might use.
Answer
MSE and SSIM - This is a possible solution, as suggested by your prof. As I assume the low quality images also have a different resolution than the good ones, remember to downsample the good ones (and not upsample the bad ones).
Image subtraction (1-norm distance) - Subtract two images -> if they are equal you'll get a black image. If they are slightly different, the non-black pixels (or the sum of the pixel intensity) can be used as a similarity index. This is actually the 1-norm distance.
Histogram distance - You can refer to this paper: https://www.cse.huji.ac.il/~werman/Papers/ECCV2010.pdf. Comparing two images' histograms might be potentially robust for your task. Check out this question too: Comparing two histograms
Embedding learning - As I see you included tensorflow, keras or pytorch as tags, let's consider deep learning. This paper came to my
mind: https://arxiv.org/pdf/1503.03832.pdf The idea is to learn a
mapping from the image space to a Euclidian space - i.e. compute an
embedding of the image. In the embedding hyperspace, images are
points. This paper learns an embedding function by minimizing the
triplet loss. The triplet loss is meant to maximize the distance
between images of different classes and minimize the distance between
images of the same class. You could train the same model on a Dataset
like ImageNet. You could augment the dataset with by lowering the
quality of the images, in order to make the model "invariant" to
difference in image quality (e.g. down-sampling followed by
up-sampling, image compression, adding noise, etc.). Once you can
compute embedding, you could compute the Euclidian distance (as a
substitute of the MSE). This might work better than using MSE/SSIM as a similarity indexes. Repo of FaceNet: https://github.com/timesler/facenet-pytorch. Another general purpose approach (not related to faces) which might help you: https://github.com/zegami/image-similarity-clustering.
Siamese networks for predicting similarity score - I am referring to this paper on face verification: http://bmvc2018.org/contents/papers/0410.pdf. The siamese network takes two images as input and outputs a value in the [0, 1]. We can interpret the output as the probability that the two images belong to the same class. You can train a model of this kind to predict 1 for image pairs of the following kind: (good quality image, artificially degraded image). To degrade the image, again, you can combine e.g. down-sampling followed by
up-sampling, image compression, adding noise, etc. Let the model predict 0 for image pairs of different classes (e.g. different images). The output of the network can e used as a similarity index.
Remark 1
These different approaches can also be combined. They all provide you with similarity indexes, so you can very easily average the outcomes.
Remark 2
If you only need to do it once, the effort you need to put in implementing and training deep models might be not justified. I would not suggest it. Still, you can consider it if you can't find any other solution and that Mac is REALLY FULL of images and a manual search is not possible.
If you look at the documentation of imgdupes you will see there is the following option:
--dry-run
dry run (do not delete any files)
So if you run imgdupes with --dry-run you will get a listing of all the duplicate images but it will not actually delete anything. You should be able to process that output to move the images around as you need.
Try similar image finder I have developed to address this problem.
There is an explanation and the algorithm there, so you can implement your own version if needed.
When you have variable sized input images and use the ImageDeserializer to resize the images, how are you supposed to deal with a mean file? Computing the mean file is easy when the input images are all the same size. Wouldn't it be better if the ImageDeserializer would be capable of compute the means?
The order in which image pre-processing steps are executed by default are:
Take a crop that has the desired image size
Subtract the mean
Hence, your mean file hence only has to contain a mean for the desired input size.
For computing the mean yourself, you will have to repeat these steps, at least to some degree. If you're on .NET, then you may want to have a look at this post where .NET image pre-processing is discussed.
I agree that it would be helpful if there was some tool to compute the mean file. I can understand though why the image deserializer does not do it automatically: You need to transform your training and test data via the same mean file. If you subtract mean from training data automatically, you'll have no way of repeating the same operation on the test data. Plus there is randomization that could make it messy in some cases, etc.
Can somebody help me, to find a library, or a detailed description of algorithm, that could embed a Digital watermark(invisible watermark, just a kind of steganography) to a jpeg/png file. But the quality of algorithm, should be great. It should be possible to extract this mark after rotation and expansion(if possible) of image.
Mark is just a key 32bytes.
I found a good site, but the algorithm are made for the NetPBM format, that is dead...
I know that there is a LSB method, but it is not stable to the expansion. Are there something better?
Changing metadata, is not suitable, because it is visible changes.
This maybe won't really be an answer, as I don't think it would be easy to give a magical, precise answer on this question.Watermarking is complex, and the best way to do it is by yourself : this will make things more hard for an attacker trying to reverse engineer your code. One could even read your question here, guess what library you used, and attack your system more easily.
Making Steganography resist to expansion in JPEG images is also very hard, because the JPEG compression is reapplied after the expansion. There are in fact a bunch of JPEG steganography algorithms. Which one you should use, depends on what exactly do you require :
Data confidentiality ?
Message presence confidentiality ?
Message coherence after JPEG changes ?
Resistance to "Known Cover" attacks (when attackers try to find the message, based on the steganographic system) ?
Resistance to "Known Message" attacks (when attackers try to find the steganographic system used, based on the message) ?
From what I know, usually, algorithm that resist to JPEG changes (picture recompression) are often really easier to attack, whereas algorithms that run the "encode" stage during the JPEG compression (after the DCT (lossy) transform, and before the Huffmann (non-lossy) transform) are more prone to resist.
Also, one key factor about steganography is scale : if you have only 32bytes of data to encode in a, say, 256*256px image, don't use an algo that can encode 512bytes of data in the same size. Either use a scalable algorithm, either use an algorithm at its efficient scale.
Also, the best way to do good steganography is to know its limitations,and to know how steganalyzers work. Try these tools, so you can understand what attackers will do to your picture.^
Now, I cannot tell you what steganographic system will be the best for you, but I can give you some indications :
jSteg - Quite old, I don't think it will resist to JPEG changes
OutGuess - Quite old too, but one of the best algorithms
F5 (and F3/F4) - More recent, good algorithm, scientifical research behind.
stegHide
I think all of these are LSB based : the encoding is done during the JPEG compression, after the DCT and Quantization. The only non LSB-based steganography system I heard of was mentionned in this research paper, however, I did not read it to the end yet, so I cannot tell if this will meet your needs.
However, I'm not sure there exists a real steganography algorithm resisting to JPEG compression, to JPEG resize and rotation, resisting to visual and statisticals attacks. Or I'm not aware of it.
Sorry for the lack of precise answer, I tried to give you what I know on the subject, as it's always better to be more informed. Sorry also for the lack of proper English, I'm French, nobody's perfect :)
Pistache is right in what he told you regarding the watermarking implementation algorithms. I will try to help you by showing one algorithm for the given requirements.
Before explaining you the algorithms first I guess that the distinction between the JPG and PNG formats should be done.
JPEG is a lossy format, i.e, the images are susceptible to compression that could remove the watermark. When you open an image for manipulation purposes and you save it, upon the writing procedure, a compression is made by using DCT filtering that removes some important components of the image.
On the other hand, PNG format is lossless, and that means that images are not susceptible to such kind of compression when stored after manipulation.
As a matter of fact, JPEG is used as a watermarking scheme attack due to its compressing characteristic that could remove the watermark if an attacker performed the compression.
Now that you know the difference between both formats, I can tell you a suitable algorithm resistant to the attacks that you mentioned.
Regarding methods to embed a watermark message for PNG files you can use the histogram embedding method. The histogram embedding method changes values on the histogram by changing the values of the neighbor bins. For example imagine that you have a PNG image in grayscale.
Therefore, you'll have only one channel for embedding and that means that you have one histogram with 256 bins. By selecting the neighbor bins x and x+1, you change the values of x and x+1 by moving the pixels with the bright x to x+1 or the other way around, so that (x/(x+1))>T for embedding a '1' or ((x+1)/x)>T for embedding a '0'.
You can repeat the same procedure for the whole histogram length and therefore you can embed in the best case up to 128bits. However this payload is less than what you asked. Therefore I suggest you to split the image into parts, for example blocks, and if you split one image into 4 components you'd be able to embed in the best case up to
512 bits which means 64 bytes.
This method although is very, susceptible to filtering and compression if applied straight in the space domain. Therefore, I suggest you to compute before the DWT of the image and use its low-frequency sub-band. This will provide you better transparency and robustness increased for the warping, resizing etc attacks and compression or filtering as well.
There are other approaches such as LPM (Log Polar Maps) but they are very complex to implement and I think for your case this approach would be fine.
I can suggest you two papers, the first is:
Watermarking digital image and video data. A state-of-the-art overview
This paper will give you some basic notions of watermarking and explain more in detail the LSB algorithm. And the second paper is:
Real-Time Compressed- Domain Video Watermarking Resistance to Geometric Distortions
This paper will explain the algorithm that I just explained now.
Cheers,
I do not know if you are considering approaches different to steganography. Instead of storing data hidden in the pixel data you could create a new data block in the JPEG file and store encripted data.
Take a look at the JPEG file structure on Wikipedia
You can create an application specific data block, using the marker 0xFF 0xEn. Doing so, any change in the image pixels do not change the information stored in the image. Moreover, many image editing software respect custom data blocks and will keep them even after image manipulation.
I've read some of posts here about png/jpg/gif but still I'm quite confused..
I've got a big header on my website :
width:850px height:380px weight:108kb
And it's jpg. A woman + gradient + some layers on top and behing her..
Do you think 108kb it's too much? I was thinking about cut it to png pieces..Would that be a bad idea? What's your suggestions?;) THX for help;]
It depends on the nature of the image, if it's a photograph, JPEG would give the highest quality/compression ratio, if it was pixel stuff like writing or clipart, or have some transparency, then it's GIF/PNG to choose (GIF,PNG8 offers one level transparency, while PNG24 offers a levelled transparency for each pixel).
When I'm confused, I usually save the picture in all three formats and decide what gives the best quality/size deal within the size limits I need, I also try to lower down JPEG quality to the level where the image is still good quality (because that varies from image to another).
Also, if it was a photograph with some writing, split it into a JPEG photograph with a transparent GIF writing overlay (because writing edges look distorted in JPEG).
So, when you are confused, give the three a try and decide, with time you'll gain experience what format suits what content the most.
Actually, 108kb for an image of that size isn't abnormal. It's even more optimal to have it in 1 image: your browser only needs to perform 1 get-request. If you break it up into multiple images, the user needs to wait longer for the site to load.
PNG probably wouldn't help you since JPG is usually much more efficient at handling gradients. PNG would be better if you had big unicolored spaces in your image.
I am using Tesseract OCR (via pytesser) and PIL (Python Image Library) for automated test of an application.
I am checking that the displayed text is ok by making a screenshot and getting the text thanks to tesseract.
I had some issues in the beginning and it seems to work better since I have increased the size of the screenshot thanks to the bicubic interpolation of PIL.
Unfortunatelly, I still have some mistakes like confusion between '0' and 'O'. I can imagine that I will have other similar issues in the future.
I would like to know if there are some techniques to prepare an image in order to help the OCR. Any idea is welcomed.
Thanks in advance
Shameless plug and disclaimer: my company packages Tesseract for use in .NET
Tesseract is an OK OCR engine. It can miss a lot and gets readily confused by non-text. The best thing you can do for it is to make sure it gets text only. The next best thing is to give it something sanely binarized (adaptive or dynamic threshold to get there) or grayscale and let it try to do binarization.
Train tesseract to recognize your font
Make image extra clean and with enough free space around characters
Profit :)
Here are few real world examples.
First image is original image (croped power meter numbers)
Second image is slightly cleaned up image in GIMP, around 50% OCR accuracy in tesseract
Third image is completely cleaned image - 100% OCR recognized without any training!
Even under the best conditions OCR variants will sneak up on you. Your best option will be to design your tests to be aware of them.
For distinguishing between 0 and O, one simple solution is to choose a font that distinguishes between both (eg: 0 has a dash or dot in its middle). Would that be acceptable in your application?
Another solution is to apply a dictionary-based step after the character-by-character analysis of the text - feeding the recognized text into some form of spell-checker or validator to differentiate between difficult characters.
For instance, a round symbol followed by other numbers is most likely to be a zero, while the same symbol followed by letters is most likely to be a capital o. It's a trivial example, but it shows how context is necessary to make a more reliable OCR system.