Reading quantization tables of jpeg files on Android - java

A previous answer provides code for reading the quantization tables of jpeg files in Java. However it uses javax.imageio.ImageIO which, according to this answer, isn't available on the Android platform.
How can I read read the quantization tables of jpeg files on Android in Kotlin or the Java subset available on Android?
edit
I have found the korlibs/korim Kotlin library which seems to read the quantization tables. But I'm still a beginner with Kotlin so I haven't yet figured out how I go from a string with the path to the JPEG file to printing out the quantization table.

I am, unfortunately, not aware of such an API for Android.
However, parsing the JPEG quantization tables aren't that hard... You could probably have a look at my JPEGSegmentUtil and QuantizationTable (originally by Helmut Dersch) classes and adapt them to Android. The mentioned code is available under BSD license.
You would need to replace the javax.imageio.ImageInputStream with a different type, preferably something that implements java.io.DataInput and has proper seek/skip, replace javax.imageio.IIOException with normal IOException, and perhaps a few more tweaks.

If you only need to read quantization tables then the simplest solution would be to implement parsing yourself. Basic sample (in C#; but Java/Kotlin implementation is pretty similar): https://stackoverflow.com/a/46162800/136138
This conversation might also be useful: JPEG file quantization table definition
If you need to do more JPEG-related operations then look at JpegKit that wraps the native libjpeg-turbo library.

Related

find compression mode of JPEG image

I was trying to open jpeg files in a Java program and noticed that neither ImageIO nor the Apache commons imaging library tools could open the images. The commons library showed me this error:
"Only sequential, baseline JPEGs are supported at the moment"
So, my image files are compressed in a way both libraries aren't able to read. I could make an ImageJ macro and transform all the images first but I would like to just use my program and not something extra.
Is there a way to find the compression mode in jpeg or even a java library that can read jpegs in several modes?
Thanks in advance
Suddenly it works with the ImageIO standard Class. Don't ask me why. Thanks for your help guys.
You need to get some tool that will allow you to dump a JPEG stream. There are a number of them out there.
What you are looking for is the start of frame marker.
FFC0 indicates baseline sequential.
FFC1 indicates extended sequential. It doesn't take much more code to do extended sequential than baseline. It is puzzling why a decoder would limit itself to baseline these day.
FFC2 is progressive.
There are others but those are the only ones you are likely to encounter and are widely supported.
You just need to find a tool that will save in baseline format. Finding one to read the other two formats is easy.

How to read/write custom PNG metadata in Android?

I have to associate a couple of text parameters (a UUID and a couple of strings representing integers) to a PNG image in a way they can follow the image when the PNG file is passed from an Android device to another through the Net. This is the typical situation in which I would use a couple of custom auxiliary chuncks (similar to EXIF fields ) to store my data inside the PNG image itself.
Maybe it is just me but the only info I was able to find about reading and writing PNG custom metadata from Java code on Android was this SO post:
Writing image metadata in Java, preferably PNG
that even offers some code (quite verbose, as usual with Java).
Those same SO post refers also to PNGJ: http://code.google.com/p/pngj/
To be honest, I would be happy to not use yet another library in this project.
Does anybody know of others ways to write and read text metadata in a PNG file in Android? Maybe a less verbose way... Maybe a way that does not require a separated library....
Any other source of information? Any tutorial? Any example?
I'm open to use a different (but equivalent) image file format, if needed (JPEG, whatever).
A solution working also on iOS and Windows 8 Phone would be a plus but it is not actually required.
I had to do something similar lately, so I had to study the subject.
Android offers you no option of manipulating png metadata. You will have to use an external library (PNGJ seems like a good option).
In my case, since I am making changes to the Android frameworks, I didn't use an external lib, but made changes to skia (the Android graphics library instead).
If you need more info on that I can give you, but I think it's irrelevant in your case.

JAI: Reading in 12 bit JPEG files

I'm brand new to Java Advanced Imaging, and the first stumbling block I've reached is the ability to read in a 12 bit, single band, greyscale JPEG file. I've seen references to it being possible with JAI, but no code or even suggestions about how it should be done. Could someone please help me out with either a helpful link or a short code snippet?
I've been using this tutorial so far, but it hasn't helped me on this issue.
Thanks.
JAI-ImageIO will register itself into Java's ImageIO api, so you should be able to use it just by having the jai-imageio jars in your classpath an calling the normal ImageIO methods, such as ImageIO.read(file).
The problem with jpeg might be that Java has a default jpeg reader in the IIORegistry, and you may have to select the right one manually using something like ImageIO.getImageReadersForFormatName().
Another thing with the more esoteric formats is that JAI ImageIO usually has two implementations - one pure Java and the other using native binary libraries, so make sure that you include the *lib-wrapper.dll (or whatever suits your particular OS) in the LD_LiBRARY_PATH or have the in the working directory of your program. The native implementation usually supports mode format variants than the pure-java one.
we usually read JPEG 12bit grayscale images using JAI + JAI Image I/O.
Additionally to previous answers, you need the native binary libraries to read JPEG 12bit, so mediaLib library is required.
You only need the "ImageRead" operation from Image I/O:
byte[] imageBytes = ...
RenderedOp readImage = JAI.create("ImageRead", new MemoryImageInputStream(imageBytes));
Whe usually read the image files from FTP, so get the byte[] and need the custom MemoryImageInputStream that wraps a byte[] into a ImageInputStream by subclassing ImageInputStreamImpl.

How can I do LZW decoding in Java?

I have a database which contains picture data stored as a binary blob. The documentation says the data is encoded using LZW. I thought that I could decode it using the Zip or GZip input streams found in the Java library, but it didn't work - I got an exception that said the format of the data is not correct.
From what I've read, the library uses DEFLATE, which is not LZW. Also, I've read about some licensing problems for using the LZW algorithm.
What can I use to decode the data? Is there a library? Do I have to implement it myself? What about the licensing problems?
I know the question is old, but I just wanted to add a great resource about LZW:
http://www.matthewflickinger.com/lab/whatsinagif/lzw_image_data.asp
It's more specifically about the use of LZW in GIF images, but it explains the compression and decompression algorithms pretty well.
Here are a couple of links:
http://www.cs.sfu.ca/CC/365/li/squeeze/LZW.html
http://u.cs.biu.ac.il/~freskom1/AlgProg1/Progs/LZW.java
http://www.codeproject.com/KB/java/lzw.aspx
And there are others.
Indeed if the images are LZW compressed TIFF files, The Java Advanced Imaging API apparently supports decoding directly (though not encoding it seems).
You can also try with 7-Zip JBinding which uses the 7zip library internally. It's quite easy to use.
I went through a surprising amount of LZW implementations before finding one that worked for my case.
UncompressedInputStream from the BioJava project worked for me, when I needed to unpack a .pax file.

Reading RAW images from Java

Canon/Nikon/other cameras save raw output of their sensor in some of their proprietary formats (.CR2, whatever). Is there any Java library designed to read them and convert into manageable BufferedImages?
I don't reqlly care here about fully customizable conversion similar to ufraw or imagemagick, rather something simple that "just works" for rendering simple previews of such images.
I've been where you are, and I feel for you. Your best bet is to use an Adobe or dcraw-based program to create thumbnails automatically. Temporary DNG files using Adobe's converter may be easier to use.
IF you insist on doing it in Java, you're about to run into a mountain of pain. RAW formats change often, have all sorts of crazy nuances and are intentionally hard to work with. Camera makers want you to use THEIR RAW conversion software, to show the camera's abilities at its best and screw Adobe. The guy behind dcraw found that some are camera manufacturers even use encryption now!
The existing Java libraries are poor -- JRawIO has improved since I last looked at it, but it supports only a fraction of the formats that dcraw does. In addition to the listed libraries, the imagero library may provide the ability to display a thumbnail for your image.
From personal experience, don't even think about writing your own RAW file reader.
I tried to do this with a very simple RAW format once (just a solid array of sensor data, 12 bits per pixel). The dcraw source translates badly to Java. You haven't seen such a nightmare of bit-fiddling ever. Took me days to debug problems with byte alignment and endian-ness.
jrawio is a plugin for Java Image I/O. With it you can read the raster data, the thumbnails and the metadata from the raw image file.
nef and cr2 already contains preview images in jpeg. just find the right offset and the right length to extract it...
Laurent Clevy # lclevy.free.fr/raw
Laurent
Unless you want to write you own file parser/loader (sounds fun imho ;) ), perhaps JMagick will help you. I haven't tried it and it might not work given your target platform since JMagick uses JNI.
UPDATE: dcraw looks like a good resource/reference

Categories

Resources