Introduction

About jpylyzer

This User Manual documents jpylyzer, a validator and feature extractor for JP2 images. JP2 is the still image format that is defined by JPEG 2000 Part 1 (ISO/IEC 15444-1). Jpylyzer was specifically created to answer the following questions that you might have about any JP2 file:

1. Is this really a JP2 and does it really conform to the format’s specifications (validation)?

2. What are the technical characteristics of this image (feature extraction)?

Starting with version 2.2, jpylyzer also supports Part 15 of the standard (ISO/IEC 15444-15, High Throughput JPEG 2000), and its JPH image format.

Validation: scope and restrictions

Since the word ‘validation’ means different things to different people, a few words about the overall scope of jpylyzer. First of all, it is important to stress that jpylyzer is not a ‘one stop solution’ that will tell you that an image is 100% perfect. What jpylyzer does is this: based on the JP2 format specification (ISO/IEC 15444-1), it parses a file. It then subjects the file’s contents to a large number of tests, each of which is based on the requirements and restrictions that are defined by the standard. If a file fails one or more tests, this implies that it does not conform to the standard, and is no valid JP2. Importantly, this presumes that jpylyzer’s tests accurately reflect the format specification, without producing false positives.

‘Valid’ means ‘probably valid’

If a file passes all tests, this is an indication that it is probably valid JP2. This (intentionally) implies a certain degree of remaining uncertainty, which is related to the following.

First of all, jpylyzer (or any other format validator for that matter) ‘validates’ a file by trying to prove that it does not conform to the standard. It cannot prove that that a file does conform to the standard.

Related to this, even though jpylyzer’s validation process is very comprehensive, it is not complete. For instance, the validation of JPEG 2000 codestreams at this moment is still somewhat limited. These limitations are discussed in detail here. Some of these limitations (e.g. optional codestream segment markers that are only minimally supported at this stage) may be taken away in upcoming versions of the tool.

No check on compressed bitstreams

One important limitation that most certainly will not be addressed in any upcoming versions is that jpylyzer does not analyse the data in the compressed bitstream segments. Doing so would involve decoding the whole image, and this is completely out of jpylyzer’s scope. As a result, it is possible that a JP2 that passes each of jpylyzer’s tests will nevertheless fail to render correctly in a viewer application.

Recommendations for use in quality assurance workflows

Because of the foregoing, a thorough JP2 quality assurance workflow should not rely on jpylyzer (or any other format validator) alone, but it should include other tests as well. Some obvious examples are:

• A rendering test that checks if a file renders at all

• Format migration workflows (e.g. TIFF to JP2) should ideally also include some comparison between source and destination images (e.g. a pixel-wise comparison)

Conversely, an image that successfully passes a rendering test or pixel-wise comparison may still contain problematic features (e.g. incorrect colour space information), so validation, rendering tests and pixel-wise comparisons are really complementary to each other.

Outline of this User Manual

We start by describing the installation process of jpylyzer for Windows and Unix-based systems. We then explain its basic usage, either as a command-line tool, or as an importable Python module. This is followed by a brief overview of the structure of the JP2 format and its ‘box’ structure, and an explanation of Jpylyzer’s output format. The final sections give a detailed description of the tests that jpylyzer performs for validation, and the properties that are reported in the output. The penultimate section does this for all ‘boxes’, except the ‘Contiguous Codestream’ box, which is given a section of its own.

Funding

The development of jpylyzer was funded by the EU FP 7 project SCAPE (SCAlable Preservation Environments). More information about this project can be found here:

https://www.scape-project.eu/

License

Jpylyzer is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this program. If not, see:

https://www.gnu.org/licenses/

On Debian systems, the complete text of the GNU Lesser General Public License version 3 can be found in:

/usr/share/common-licenses/LGPL-3

Installation and set-up

Obtaining the software

To obtain the latest version of the software please use the download links at the jpylyzer homepage:

https://jpylyzer.openpreservation.org/

You have three options:

1. Install the software with the Pip package manager. This works on all platforms (Windows, Linux, Mac, etc.), but you need to have the Python interpreter available on your system. Jpylyzer is compatible with Python 3.2 and more recent.

2. Alternatively, for Windows users there is also a set of stand-alone binaries¹. These allow you to run jpylyzer as an executable Windows application, without any need for installing Python. This option is particularly useful for Windows users who cannot (or don’t want to) install software on their system.

3. For Linux users Debian packages are available.

These options are described in the following sub-sections.

Installation with Pip (Linux/Unix, Windows, Mac OS X)

General installation procedure

First make sure you have a recent version of pip. Then install jpylyzer with the following command:

pip install jpylyzer

Single user installation (Linux)

On most Linux systems the above command needs to be run as super user (see below). If you don’t want this use the below command for a single-user install:

pip install jpylyzer --user

This will install the software to the .local folder (hidden by default!) in your home directory (~/.local). Next try to run jpylyzer by entering:

jpylyzer

Most likely this will result in:

jpylyzer: command not found

If this happens, add the directory ~/.local/bin (which is where the jpylyzer command-line tool is installed) to the PATH environment variable (you only need to do this once). To do this, locate the (hidden) file .profile in you home directory (~/), and open it in a text editor. Then add the following lines at the end of the file:

# set PATH so it includes the user's .local bin if it exists
if [ -d "$HOME/.local/bin" ] ; then
    PATH="$HOME/.local/bin:$PATH"
fi

Save the file, log out of your session and then log in again. Open a command terminal and type:

jpylyzer

If all went well you now see this:

usage: jpylyzer [-h] [--format FMT] [--mix {1,2}] [--nopretty]
          [--nullxml] [--recurse] [--packetmarkers] [--verbose]
          [--version] jp2In [jp2In ...]
jpylyzer: error: the following arguments are required: jp2In

Which means that the installation was successful!

Global installation (Linux)

Simply enter:

sudo -H pip install jpylyzer

No further configuration is needed in this case.

Note on pre-releases

The above command lines will only install stable versions of jpylyzer. In order to install the latest pre-release, add the --pre switch. For example:

sudo -H pip install jpylyzer --pre

Installation of Windows binaries (Windows only)

Download the binary using the link on the jpylyzer homepage. Unzip the contents of this file to an empty folder on your PC. Jpylyzer should now be ready for use.

Testing the installation

To test your installation, open a Command Prompt (‘DOS prompt’) and type:

%jpylyzerPath%\jpylyzer

In the above command, replace %jpylyzerPath% with the full path to the jpylyzer installation directory (i.e. the directory that contains ‘jpylyzer.exe’ and its associated files). For example, if you extracted the files to directory c:\tools\jpylyzer, the command would become:

c:\tools\jpylyzer\jpylyzer

Executing this command should result in the following screen output:

usage: jpylyzer [-h] [--format FMT] [--mix {1,2}] [--nopretty]
          [--nullxml] [--recurse] [--packetmarkers] [--verbose]
          [--version] jp2In [jp2In ...]
jpylyzer: error: the following arguments are required: jp2In

Running jpylyzer without typing the full path

Optionally, you may also want to add the full path of the jpylyzer installation directory to the Windows ’Path’ environment variable. Doing so allows you to run jpylyzer from any directory on your PC without having to type the full path. In Windows 7 you can do this by selecting ‘settings’ from the ‘Start’ menu; then go to ‘control panel’/’system’ and go to the ‘advanced’ tab. Click on the ‘environment variables’ button. Finally, locate the ‘Path’ variable in the ‘system variables’ window, click on ‘Edit’ and add the full jpylyzer path (this requires local Administrator privileges). The settings take effect on any newly opened command prompt.

Installation of Debian packages (Ubuntu/Linux)

For Linux, Debian packages of jpylyzer exist. To install, simply download the .deb file, double-click on it and select Install Package. Alternatively you can also do this in the command terminal by typing:

sudo dpkg -i opf-jpylyzer_2.2.0_all.deb

In both cases you need to have administrative privileges.

For Ubuntu and Debian alternative packages are available in the official release channels. To install simply run the following commands:

sudo apt-get update
sudo apt-get install python-jpylyzer

Using jpylyzer

Overview

This section describes the general use of jpylyzer. The first sub-sections cover the use of jpylyzer as a command-line tool and as an importable Python module.

Command-line usage

This section explains jpylyzer’s general command-line interface. For the sake of brevity, full paths to jpylyzer are omitted. This means that, depending on your system and settings, you may have to substitute each occurrence of ‘jpylyzer’ with its full path, the corresponding Windows binary, or a combination of both. The following examples illustrate this:

Furthermore, command line arguments that are given between square brackets (example: [-h]) are optional.

Synopsis

Jpylyzer can be invoked using the following command-line arguments:

usage: jpylyzer [-h] [--format FMT] [--mix {1,2}] [--nopretty]
          [--nullxml] [--recurse] [--packetmarkers] [--verbose]
          [--version] jp2In [jp2In ...]

Positional arguments

Optional arguments

Note that the input can either be a single image, a space-separated sequence of images, a pathname expression that includes multiple images, or any combination of the above. For example, the following command will process one single image:

jpylyzer rubbish.jp2

The next example shows how to process all files with a ‘jp2’ extension in the current directory:

jpylyzer *.jp2

Note that on Unix/Linux based systems pathname expressions may not work properly unless you wrap them in quotation marks:

jpylyzer "*.jp2"

Output redirection

All output (except warning and system error messages) is directed to the standard output device (stdout). By default this is the console screen. Use your platform’s standard output redirection operators to redirect output to a file. The most common situation will be to redirect the output of one invocation of jpylyzer to an XML file, which can be done with the ‘>’ operator (both under Windows and Linux):

jpylyzer jp2In > outputFile

E.g. the following command will run jpylyzer on image ‘rubbish.jp2’ and redirects the output to file ‘rubbish.xml’:

jpylyzer rubbish.jp2 > rubbish.xml

The format of the XML output is described here.

‘format’ option

By default, jpylyzer validates against the JP2 format specification. Starting with version 2.0, jpylyzer can also validate raw JPEG 2000 codestreams that are not wrapped inside a JP2 container. For codestream validation, use the --format option with value j2c, e.g.:

jpylyzer --format j2c rubbish.j2c > rubbish.xml

Starting with version 2.2, jpylyzer also supports JPEG 2000 Part 15 (High Throughput JPEG 2000) and its JPH container format. To validate a JPH file, use the --format option with value jph:

jpylyzer --format jph rubbish.jph > rubbish.xml

And use --format with value jhc to validate a raw High Throughput codestream:

jpylyzer --format jhc rubbish.jhc > rubbish.xml

‘mix’ option

When this option is used, jpylyzer reports additional output in NISO MIX format. This option takes one argument that defines whether MIX 1.0 or MIX 2.0 is used. For example, the following command will result in MIX 2.0 output:

jpylyzer --mix 2 rubbish.jp2 > rubbish.xml

The MIX output is wrapped inside a file/propertiesExtension element. Note that MIX output is only written for files that are valid JP2 or JPH (files that are not valid result in an empty propertiesExtension element). Also, the --mix option is ignored if --format is set to j2c.

‘recurse’ option

If the --recurse option is used, jpylyzer will recursively traverse all subdirectories of a filepath expression. E.g:

jpylyzer /home/myJP2s/*.jp2 > rubbish.xml

In this case jpylyzer analyses all files that have a .jp2 extension in directory /home/myJP2s and all its subdirectories.

‘packetmarkers’ option

When this option is enabled, jpylyzer will report the properties of the following packet-level codestream markers:

• PLM (packet length, main header) marker
• PPM (packed packet headers, main header) marker
• PLT (packet length, tile-part header) marker
• PPT (packed packet headers, tile-part header) marker

By default these are excluded from the output, in order to prevent excessive output size.

‘nullxml’ option

The nullxml option was added to enable extraction of XML content that is terminated by a null-byte. By default jpylyzer doesn’t report the XML in that case, because it throws an exception in the XML parser. Apparently some old versions of the Kakadu demo applications would erroneously add a null-byte to embedded XML, so this option can be used to force extraction for images that are affected by this.

User warnings

Under the following conditions jpylyzer will report a user warning:

No images to check

If there are no input images to check (typically because the value of jp2In refers to a non-existent file), the following warning message is shown:

User warning: no images to check!

Unsupported box

In some cases you will see the following warning message:

User warning: ignoring 'boxName' (validator function not yet implemented)

The reason for this: a JP2 file is made up of units that are called ‘boxes’. This is explained in more detail here. Each ‘box’ has its own dedicated validator function. At this stage validator functions are still missing for a small number of (optional) boxes. Jpylyzer will display the above warning message if it encounters a (yet) unsupported box. Any unsupported boxes are simply ignored, and the remainder of the file will be analyzed (and validated) normally.

Error while processing a file

In rare cases you may come across one of the following messages:

User warning: memory error (file size too large)

Memory errors may occur for (very) large images. If you get this warning, try using a machine with more RAM. Also, a machine’s chip architecture and the operating system may put constraints on the amount of memory that can be allocated.

The following warning indicates an input error:

User warning: I/O error (cannot open file)

Finally, the following messages most likely indicate a jpylyzer bug:

User warning:runtime error (please report to developers)

and:

User warning: unknown error (please report to developers)

If you ever run into either of these two errors, please get in touch with the jpylyzer developers. The easiest way to do this is to create a new issue at:

https://github.com/openpreserve/jpylyzer/issues

Unknown box

Occasionally, you may see this warning message:

User warning: ignoring unknown box

This happens if jpylyzer encounters a box that is not defined by JPEG 2000 Part 1. It should be noted that, to a large extent, JPEG 2000 Part 1 permits the presence of boxes that are defined outside the standard. Again, jpylyzer will simply ignore these and process all other boxes normally.

All user warnings are printed to the standard error device (typically the console screen). File-level warnings are also written to the warnings output element.

Using jpylyzer as a Python module

Instead of using jpylyzer from the command-line, you can also import it as a module in your own Python programs. To do so, install jpylyzer with pip. Then import jpylyzer into your code by adding:

from jpylyzer import jpylyzer

Subsequently you can call any function that is defined in jpylyzer.py. In practice you will most likely only need the checkOneFile function. The following minimal script shows how this works:

#! /usr/bin/env python3

from jpylyzer import jpylyzer

# Define JP2
myFile = "/home/johan/jpylyzer-test-files/aware.jp2"

# Analyse with jpylyzer, result to Element object
myResult = jpylyzer.checkOneFile(myFile)

# Return image height value
imageHeight = myResult.findtext('./properties/jp2HeaderBox/imageHeaderBox/height')
print(imageHeight)

Here, myResult is an Element object that can either be used directly, or converted to XML using the ElementTree module². The structure of the element object follows the XML output that described here.

You may use the following optional arguments (which correspond to the command-line options explained above):

• validationFormat - sets the validation format. Values: ‘jp2’ (default), ‘jph’, ‘j2c’ or ‘jhc’.
• verboseFlag - report test results in verbose format. Values: False (default) or True.
• packetmarkersFlag - report packet-level codestream markers. Values: False (default) or True.
• nullxmlFlag - extract null-terminated XML content from XML and UUID boxes. Values: False (default) or True.
• mixFlag - report additional output in NISO MIX format (version 1.0 or 2.0). Values: 0 (default), 1 or 2.

As an example, for validating a raw JPEG 2000 codestream, call the checkOneFile function with the additional validationFormat argument, and set it to j2c:

# Define Codestream
myFile = "/home/johan/jpylyzer-test-files/rubbish.j2c"

# Analyse with jpylyzer, result to Element object
myResult = jpylyzer.checkOneFile(myFile, 'j2c')

Java integration

It is possible to integrate jpylyzer into Java applications³. A test class that shows how this works is included in the source repo here. This requires Jython. Note that you may run into performance issues with (very) large images in this case, as Jython does not support memory mapping, so make sure you’ve got plenty of RAM available.

Structure of a JP2 file

Scope of this section

This section gives a brief overview of the JP2 file format. A basic understanding of the general structure of JP2 is helpful for appreciating how jpylyzer performs its validation. It will also make it easier to understand jpylyzer‘s extracted properties, as these are reported as a hierarchical tree that corresponds to the internal structure of JP2.

For an exhaustive description of every detail of the format you are advised to consult Annex I (‘JP2 file format syntax’) and Annex A (‘Codestream syntax’) of ISO/IEC 15444-1.

General format structure

At the highest level, a JP2 file is made up of a collection of boxes. A box can be thought of as the fundamental building block of the format. Some boxes (‘superboxes’) are containers for other boxes. The Figure below gives an overview of the top-level boxes in a JP2 file.

A number of things here are noteworthy to point out:

• Some of these boxes are required, whereas others (indicated with dashed lines in the Figure) are optional.

• The order in which the boxes appear in the file is subject to some constraints (e.g. the first box in a JP2 must always be a ‘Signature’ box, followed by a ‘File Type’ box).

• Some boxes may have multiple instances (e.g. ‘Contiguous Codestream’ box), whereas others must be unique (e.g. ‘JP2 Header’ box).

More specific details can be found in the standard. The important thing here is that requirements like the above are something that should be verified by a validator, and this is exactly what jpylyzer does at the highest level of its validation procedure.

General structure of a box

All boxes are defined by a generic binary structure, which is illustrated by the following Figure:

Most boxes are made up of the following three components:

1. A fixed-length ‘box length’ field that indicates the total size of the box (in bytes).

2. A fixed-length ‘box type’ field which specifies the type of information that can be found in this box

3. The box contents, which contains the actual information within the box. Its internal format depends on the box type. The box contents of a ‘superbox’ will contain its child boxes (which can be parsed recursively).

In some cases a box will also contain an ‘extended box length field’. This field is needed if the size of a box exceeds 2³²-1 bytes, which is the maximum value that can be stored in the 4-byte ‘box length’ field.

Defined boxes in JP2

The following Table (taken from Table I.2 in ISO/IEC 15444-1, with minor modifications) lists all boxes that are defined in the standard. Addition signs in the ‘box name’ column indicate boxes that are children of a ‘superbox’.

A JP2 file may contain boxes that are not defined by the standard. Such boxes are simply skipped and ignored by conforming reader applications.

Output format

This section explains jpylyzer’s output format.

Overview

Jpylyzer generates its output in XML format, which is defined by the schema that can be found here. The following Figure shows the output structure:

The root element (jpylyzer) contains the following child elements:

1. one toolInfo element, which contains information about jpylyzer

2. one or more file elements, each of which contain information about about the analysed files

The XML output is pretty-printed. You can use the --nopretty switch to disable pretty-printing (this produces smaller files and may give a slightly better performance).

toolInfo element

This element holds information about jpylyzer. Currently it contains the following sub-elements:

• toolName: name of the analysis tool (i.e. jpylyzer or jpylyzer.exe, depending on the platform used)

• toolVersion: version of jpylyzer (jpylyzer uses a date versioning scheme)

file element

The file element contains the following sub-elements:

1. fileInfo: general information about the analysed file

2. statusInfo: information about the status of jpylyzer’s validation attempt

3. isValid: outcome of the validation

4. tests: outcome of the individual tests that are part of the validation process (organised by box)

5. properties: image properties (organised by box)

6. propertiesExtension: wrapper element for NISO MIX output (only if the --mix option is used)

7. warnings: reported warnings

fileInfo element

This element holds general information about the analysed file. Currently it contains the following sub-elements:

• filename: name of the analysed file without its path (e.g. “rubbish.jp2”)

• filePath: name of the analysed file, including its full absolute path (e.g. “d:\data\images\rubbish.jp2”)

• fileSizeInBytes: file size in bytes

• fileLastModified: last modified date and time

statusInfo element

This element holds general information about about the status of jpylyzer’s attempt at validating a file. It tells you whether the validation process could be completed without any internal jpylyzer errors. It contains the following sub-elements:

• success: a Boolean flag that indicates whether the validation attempt completed normally (“True”) or not (“False”). A value of “False” indicates an internal error that prevented jpylyzer from validating the file.

• failureMessage: if the validation attempt failed (value of success equals “False”), this field gives further details about the reason of the failure. Examples are:

  • memory error (file size too large)
  • runtime error (please report to developers)
  • unknown error (please report to developers)

isValid element

This element contains the results of the validation. If a file passed all the tests (i.e. all tests returned “True”, see here) it is most likely valid, and the value of isValid will be “True”. Its value is “False” otherwise. The element has a format attribute, which defines the validation format (set by the --format command-line option). The format attribute can have the following values:

• “jp2” (JP2 validation)
• “j2c” (raw codestream validation)

tests element

This element is reserved to hold the outcomes of all the individual tests that jpylyzer performs to assess whether a file is valid JP2. The results are organised in a hierarchical tree that corresponds to JP2’s box structure. Each individual test can have two values:

• “True” if a file passed the test.

• “False” if a file failed the test.

If a file passed all tests, this is an indication that it is most likely valid JP2. In that case, the isValid element has a value of “True” (and “False” in all other cases). These tests are all explained here and here.

Default and verbose reporting of test results

By default, jpylyzer only reports any tests that failed (i.e. returned “False”), including the corresponding part of the box structure. For a valid JP2 the tests element will be empty. If the –verbose flag is used, the results of all tests are included (including those that returned “True”)⁴.

properties element

This element contains the extracted image properties, which are organised in a hierarchical tree that corresponds to JP2’s box structure. See here and here for a description of the reported properties.

propertiesExtension element

This optional element is reserved for output in alternative formats. Currently it is used to wrap output in NISO MIX format if the --mix option is used. See the MIX documentation for a description of the reported elements.

warnings element

This element is reserved for reporting any warnings that are not directly related to the validation process. Examples are:

• Jpylyzer encountered an unknown box type while parsing a file.
• Reading of a file failed because of system limitations.

Each warning is wrapped in a ‘warning’ element. The results are organised in a hierarchical tree that corresponds to JP2’s box structure.

JP2: box by box

The following two sections provide a detailed explanation of jpylyzer’s functionality and its output. In particular, the following two aspects are addressed:

1. The reported properties

2. The tests that jpylyzer performs to establish the validity of a file.

About the properties and tests trees

The ‘properties’ element in jpylyzer’s output holds a hierarchical tree structure that contains all extracted properties. The ‘tests’ tree follows the same structure. The hierarchy reflects JP2’s box structure (explained here): each box is represented by a corresponding output element that contains the corresponding property entries. If a box is a superbox, the output element will contain child elements for each child box. For some boxes, the output contains further sub-elements. This applies in particular to the Contiguous Codestream box, since its contents are more complex than any of the other boxes. Also, if a Colour Specification box contains an embedded ICC profile, the properties of the ICC profile are stored in a separate sub-element. In addition to this, one ‘property’ that is reported by jpylyzer (the compression ratio) is not actually extracted from any particular box. Instead, it is calculated from the file size and some properties from the Header boxes. As a result, it is reported separately in the root of the properties tree.

Naming of properties

The naming of the reported properties largely follows the standard (ISO/IEC 15444-1). Some minor differences follow from the fact that the standard does have any consistent use of text case, whereas jpylyzer uses lower camel case. In addition, some parameters in the standard are compound units that aggregate a number of Boolean ‘switches’, where no names are provided for each individual switch. An example of this is the Scod (coding style) parameter in the codestream header, which contains three switches that define the use of precincts, start-of-packet markers and end-of-packet markers. For cases like these jpylyzer uses its own (largely self-descriptive) names (which are all documented in these sections).

JPEG 2000 Signature box

This box contains information that allows identification of the file as being part of the JPEG 2000 family of file formats.

Element name

signatureBox

Reported properties

None (box only holds JPEG 2000 signature, which includes non-printable characters)

Tests

File Type box

This box specifies file type, version and compatibility information, including specifying if this file is a conforming JP2 file or if it can be read by a conforming JP2 reader.

Element name

fileTypeBox

Reported properties

Tests

JP2 Header box (superbox)

This box is a superbox that holds a series of boxes that contain header-type information about the file.

Element name

jp2HeaderBox

Reported properties

Since this is a superbox, it contains a number of child boxes. These are represented as child elements in the properties tree:

Tests

Image Header box (child of JP2 Header box)

This box specifies the size of the image and other related fields.

Element name

imageHeaderBox

Reported properties

Tests

Bits Per Component box (child of JP2 Header box)

This (optional) box specifies the bit depth of the components in the file in cases where the bit depth is not constant across all components.

Element name

bitsPerComponentBox

Reported properties

Tests

Colour Specification box (child of JP2 Header box)

This box specifies the colourspace of the image.

Element name

colourSpecificationBox

Reported properties

Reported properties of ICC profiles

If the colour specification box contains an embedded ICC profile, jpylyzer will also report the following properties (which are all grouped in an “icc” sub-element in the properties tree). An exhaustive explanation of these properties is given in the ICC specification (ISO 15076-1 / ICC.1:2004-10). Note that jpylyzer does not validate embedded ICC profiles (even though it does check if a specific ICC profile is allowed in JP2)!

Tests

Palette box (child of JP2 Header box)

This (optional) box specifies the palette which maps a single component in index space to a multiple-component image.

Element name

paletteBox

Reported properties

Tests

Component Mapping box (child of JP2 Header box)

This (optional) box specifies the mapping between a palette and codestream components.

Element name

componentMappingBox

Reported properties

Tests

Channel Definition box (child of JP2 Header box)

This (optional) box specifies the type and ordering of the components within the codestream, as well as those created by the application of a palette.

Element name

channelDefinitionBox

Reported properties

Tests

Resolution box (child of JP2 Header box, superbox)

This (optional) box contains the grid resolution.

Element name

resolutionBox

Reported properties

Since this is a superbox, it contains one or two child boxes. These are represented as child elements in the properties tree:

Tests

Capture Resolution box (child of Resolution box)

This (optional) box specifies the grid resolution at which the image was captured.

Element name

captureResolutionBox

Reported properties

Resolution information in this box is stored as a set of vertical and horizontal numerators, denominators and exponents. Jpylyzer also reports the corresponding grid resolutions in pixels per meter and pixels per inch, which are calculated from these values.

Tests

Default Display Resolution box (child of Resolution box)

This (optional) box specifies the default grid resolution at which the image should be displayed.

Element name

displayResolutionBox

Reported properties

Resolution information in this box is stored as a set of vertical and horizontal numerators, denominators and exponents. Jpylyzer also reports the corresponding grid resolutions in pixels per meter and pixels per inch, which are calculated from these values.

Tests

Contiguous Codestream box

This box contains the codestream. See here.

Intellectual Property box

This (optional) box contains intellectual property information about the image. The JP2 format specification (ISO/IEC 15444-1) does not provide any specific information about this box, other than stating that “the definition of the format of [its] contents […] is reserved for ISO”. As a result, jpylyzer does not currently include a validator function for this box, which is now simply ignored. Jpylyzer will display a user warning message in that case.

XML box

This (optional) box contains XML formatted information.

Element name

xmlBox

Reported properties

If the contents of this box are well-formed XML (see ‘tests’ below), the ‘xmlBox’ element in the properties tree will contain the contents of the XML box. Note that, depending on the character encoding of the original XML, it may contain characters that are not allowed in the encoding that is used for jpylyzer’s output. Any such characters will be represented by numerical entity references in the output. If the box contents are not well-formed XML, no properties are reported for this box.

Tests

Note that jpylyzer does not check whether the XML is valid, as this is not required by the standard. Besides, doing so would make jpylyzer significantly slower for XML that contains references to external schemas and DTDs.

UUID box

This (optional) box contains additional (binary) information, which may be vendor-specific. Some applications (e.g. Kakadu and ExifTool) also use this box for storing XMP metadata (see Section 1.1.4 in Part 3 of the XMP specification¹⁹).

Element name

uuidBox

Reported properties

If the value of uuid indicates the presence of XMP metadata and the contents of this box are well-formed XML, (see ‘tests’ below), the ‘uuidBox’ element in the properties tree will contain the XMP data. Note that, depending on the character encoding of the original XML, it may contain characters that are not allowed in the encoding that is used for jpylyzer’s output. Any such characters will be represented by numerical entity references in the output. In all other cases, the ‘uuidBox’ element will contain a standard string representation the of UUID.

Note that except for the XMP case, jpylyzer will not be able to report any information on the actual contents of this box, since it is defined outside of the scope of JPEG 2000.

Tests

UUID Info box (superbox)

This (optional) box contains additional information associated with a UUID.

Element name

uuidInfoBox

Reported properties

This is a superbox which contains two child boxes. These are represented as child elements in the properties tree:

Tests

UUID List box (child of UUID Info box)

This (optional) box specifies a list of UUIDs.

Element name

uuidListBox

Reported properties

Tests

Data Entry URL box (child of UUID Info box)

This (optional) box specifies a URL.

Element name

urlBox

Reported properties

Tests

Unknown box

An image may contain boxes that are not defined by ISO/IEC 15444-1. Although jpylyzer ignores such boxes, it will report some minimal info that will allow interested users to identify them to a limited extent.

Element name

unknownBox

Reported properties

Top-level tests and properties

This sub-section describes the tests and output for the top file level.

Element name

properties

Reported properties

The metrics that are listed here are not ‘properties’ in a strict sense; instead they are secondary or derived metrics that are calculated by combining information from different parts / boxes of the file.

The compression ratio is calculated as the ratio between the size of the uncompressed image data and the actual file size:

$\mathrm{compressionRatio}=\frac{\mathrm{sizeUncompressed}}{\mathrm{sizeCompressed}}$

Here, sizeCompressed is simply the file size (fileSizeInBytes in output file’s ‘fileInfo’ element). The uncompressed size (in bytes) can be calculated by multiplying the number of bytes per pixel by the total number of pixels:

$\mathrm{sizeUncompressed}=\frac{1}{8}\sum _{i=1}^{\mathrm{nC}}{\mathrm{bPCDepth}}_i•\mathrm{height}•\mathrm{width}$

With:

nC

number of image components (from Image Header box)

i

component index

bPCDepth_i

bits per component for component i (from Image Header box or Bits Per Component box)

height

image height (from Image Header box)

width

image width (from Image Header box)

In addition, the root of the properties tree contains the elements for all top-level boxes:

Tests

The tests that jpylyzer performs at the root level fall in either of the following two categories:

1. Tests for the presence of required top-level boxes, the order in which they appear and restrictions on the number of instances for specific boxes

2. Tests for consistency of information in different parts of the file. In particular, a lot of the information in the Image Header box is redundant with information in the codestream header, and jpylyzer performs a number of tests to verify the consistency between these two.

Contiguous Codestream box

General codestream structure

The Contiguous Codestream box holds the JPEG 2000 codestream, which contains the actual image data in a JP2.

Markers and marker segments

A codestream is made up of a number of functional entities which are called markers and marker segments. A marker is essentially a 2-byte delimiter that delineates the start or end position of a functional entity. A marker segment is the combination of a marker and a set of associated parameters (segment parameters). However, not every marker has any associated parameters.

General structure of the codestream

The codestream is made up of a number of components. The Figure below gives an overview.

From top to bottom, the Figure shows the following components:

1. A start of codestream (SOC) marker, which indicates the start of the codestream

2. A main codestream header (which includes a number of header marker segments)

3. A sequence of one or more tile parts. Each tile part consists of the following components:

   1. A start of tile-part (SOT) marker segment, which indicates the start of a tile part, and which also contains index information of the tile part and its associated tile

   2. Optionally this may be followed by one or more additional tile-part header marker segments

   3. A start of data (SOD) marker that indicates the start of the bitstream for the current tile part

   4. The bitstream

4. An ‘end of codestream’ (EOC) marker that indicates the end of the codestream.

Limitations of codestream validation

It is important to stress here that jpylyzer currently doesn’t support the full set of marker segments that can occur in a codestream. As a result, the validation of codestreams is somewhat limited. These limitations are discussed in this sub-section.

Main codestream header

Annex A of ISO/IEC 15444-1 lists a total of 13 marker segments that can occur in the main codestream header. Most of these are optional. The current version of jpylyzer only offers full support (i.e. reads and validates) for the following main header marker segments (which includes all the required ones):

• Start of codestream (SOC) marker segment (required)

• Image and tile size (SIZ) marker segment (required)

• Coding style default (COD) marker segment (required)

• Coding style component (COC) marker segment (optional)

• Region-of-interest (RGN) marker segment (optional)

• Quantization default (QCD) marker segment (required)

• Quantization component (QCC) marker segment (optional)

• Progression order change (POC) marker segment (optional)

• Component registration (CRG) marker segment (optional)

• Comment (COM) marker segment (optional)

In addition the codestream header may also contain any of the following marker segments, which are all optional:

• Packet length, main header (PLM) marker segment (optional) ^*

• Packed packet headers, main header (PPM) marker segment (optional) ^*

• Tile-part lengths (TLM) marker segment (optional) ^*

The optional markers that are marked with an asterisk above are only minimally supported at this stage.

Tile parts

The tile part validation has similar limitations. The standard lists 11 marker segments that can occur in the tile part header. Currently, jpylyzer only fully supports the following ones:

• Start of tile part (SOT) marker segment (required)

• Coding style default (COD) marker segment (optional)

• Coding style component (COC) marker segment (optional)

• Region-of-interest (RGN) marker segment (optional)

• Quantization default (QCD) marker segment (optional)

• Quantization component (QCC) marker segment (optional)

• Progression order change (POC) marker segment (optional)

• Comment (COM) marker segment (optional)

• Start of data (SOD) marker segment (required)

In addition the following optional marker segments may also occur:

• Packet length, tile-part header (PLT) marker segment (optional) ^*

• Packed packet headers, tile-part header (PPT) marker segment (optional) ^*

The optional markers that are marked with an asterisk above are only minimally supported at this stage: if jpylyzer encounters them, it will include the corresponding element in the properties element of the output. However, jpylyzer does not analyse the contents of these marker segments, which means that the respective elements in the output will be empty.

Bit streams

In addition to the above limitations, jpylyzer can not be used to establish whether the data in the bitstream are correct (this would require decoding the compressed image data, which is completely out of jpylyzer’s scope)²⁰. As a result, if jpylyzer is used as part of a quality assurance workflow, it is recommended to also include an additional check on the image contents²¹. Also, jpylyzer does not perform any checks on marker segments within the bit-stream: start-of packet (SOP) and end-of-packet (EPH) markers.

Detection of incomplete or truncated codestreams

A JP2’s tile part header contains information that makes it possible to detect incomplete and truncated codestreams in most cases. Depending on the encoder software used, this method may fail for images that only contain one single tile part (i.e. images that do not contain tiling).

Current limitations of comment extraction

Both the codestream header and the tile part header can contain comment marker segments, which are used for embedding arbitrary binary data or text. Jpylyzer will extract the contents of any comments that are text.

Structure of reported output

The Figure below illustrates the structure of jpylyzer’s codestream-level output.

At the top level, the SIZ, COD, QCD and COM marker segments are each represented as individual sub elements. The tile part properties are nested in a tileParts element, where each individual tile part is represented as a separate tilePart sub element.

Contiguous Codestream box

Element name

contiguousCodestreamBox

Reported properties

The reported properties for this box are organised into a number groups, which are represented as child elements in the properties tree:

Note that the plm and ppm elements are only written if the --packetmarkers option is used. The number of PLM and PPM markers is given by the following 2 derived properties (these are always reported, irrespective of --packetmarkers):

Tests

Image and tile size (SIZ) marker segment (child of Contiguous Codestream box)

Element name

siz

Reported properties

Tests

Coding style default (COD) marker segment

Element name

cod

Reported properties

Tests

Coding style component (COC) marker segment

Element name

coc

Reported properties

Tests

Region-of-interest (RGN) marker segment

Element name

rgn

Reported properties

Tests

Quantization default (QCD) marker segment

Element name

qcd

Reported properties

Tests

Quantization component (QCC) marker segment

Element name

qcc

Reported properties

Tests

Progression order change (POC) marker segment

Element name

poc

Reported properties

Tests

Component registration (CRG) marker segment

Element name

crg

Reported properties

Tests

Comment (COM) marker segment

Element name

com

Reported properties

Tests

Extended capabilities (CAP) marker segment

This marker segment is defined in ISO/IEC 15444-2, and is used by ISO/IEC 15444-15 (as well as other parts of the standard).

Element name

cap

Reported properties

Tests

Profile (PRF) marker segment

Element name

prf

Reported properties

Tests

Corresponding profile (CPF) marker segment

Element name

cpf

Reported properties

Tests

Tile part (child of Contiguous Codestream box)

Tile-part level properties and tests. This is not a box or a marker segment!

Element name

tilePart (child of tileParts)

Reported properties

Each tile part element can contain a number of child elements:

Note that the plt and ppt elements are only written if the --packetmarkers option is used. The number of PLT and PPT markers is given by the following 2 derived properties (these are always reported, irrespective of --packetmarkers):

Tests

Start of tile part (SOT) marker segment (child of tile part)

Element name

sot

Reported properties

Tests

The following marker segments are only minimally supported: jpylyzer will report their presence in the properties element, but it does not perform any further tests or analyses. This may change in upcoming versions of the software.

Tile-part lengths (TLM) marker segment

Element name

tlm

Reported properties

Tests

Packet length, main header (PLM) marker segment

Element name

plm

Reported properties²³

Tests

Packed packet headers, main header (PPM) marker segment

Element name

ppm

Reported properties²⁴

Tests

Packet length, tile-part header (PLT) marker segment

Element name

plt

Reported properties²⁵

Tests

Packed packet headers, tile-part header (PPT) marker segment

Element name

ppt

Reported properties²⁶

Tests

References

ICC. Specification ICC.1:1998-09 – File Format for Color Profiles. International Color Consortium, 1998. https://www.color.org/ICC-1_1998-09.pdf.

ISO/IEC. Information technology — JPEG 2000 image coding system: Core coding system. ISO/IEC 15444-1, Second edition. Geneva: ISO/IEC, 2004a. https://www.jpeg.org/public/15444-1annexi.pdf (“Annex I: JP2 file format syntax” only).

ISO/IEC. Information technology — JPEG 2000 image coding system: Extensions. ISO/IEC 15444-2, First edition. Geneva: ISO/IEC, 2004b. https://www.jpeg.org/public/15444-2annexm.pdf (“Annex M: JPX extended file format syntax” only).

Leach, P., Mealling, M. & Salz, R. A Universally Unique IDentifier (UUID) URN namespace. Memo, IETF. https://tools.ietf.org/html/rfc4122.html.