Following the most recent book edition, the associated imagingbook software with new features and completely restructured is now available. This new release is based on Java Language Level 9 (using modules) and all components are now available as Maven artifacts (though the use of Maven is not required). It also includes a set of JAR-packaged sample images and most of the contained plugins are set up to load a suitable demo image automatically if no other image is currently open in the ImageJ workbench.

New features include:

• A completely new implementation of MSER feature detection and the underlying component tree generation, analogous (though complementary in nature) to the existing SIFT implementation.
• Implementation of additional corner detection schemes, such as the MOPS and Shi-Tomasi algorithms, with improved sub-pixel position refinement.
• Implementations of piecewise-linear warping and log-polar transformation with interactive plugins.
• Additional methods for fitting straight lines, circles and ellipses (using both algebraic and geometric fits) to 2D point sets.
• Generic framework for 2D RANSAC primitive detection, currently implemented for straight lines, circles and ellipses.
• Introduction of “parameter bundles”, i.e., a group of parameters for a particular operator. Parameter bundles are useful when operators require a large number of parameters which would otherwise lead to cluttered constructor signatures. Parameter bundles may also be passed as a whole to ImageJ’s GenericDialog to be modified by the user. Displayed descriptions and number formats are controlled by annotations.
• Automatic compilation of plugins.config files for ImageJ plugins sets at build time. Plugin names and their placement in the menu structure are controlled by Java annotations at package and class levels.
• Simplified graphic overlay generation from AWT Shape instances with the new ShapeOverlayAdapter class, which takes care of pixel-centered positioning as well as color and stroke controls.
• Introduction of “named resources” (mostly images) for added compile-time safety when loading resources and automatic testing.
• Numerous ImageJ plugins for accessing and experimenting with the imagingbook library features.
• Modular, Maven-based project structure. All components are now available as artifacts on Maven Central.

MSER feature detection example

For more details and downloads see the associated repository on GitHub.

A new relase of the imagingbook-common library has been released today. This version fixes several bugs, mostly releated to the ImageAccessor and associated classes. The new JAR files can be downloaded from this location:

https://github.com/imagingbook/imagingbook-jars

In the meantime the imagingbook source code repository has been moved to GitHub for technical reasons. An official announcement will follow shortly.

The color quantization code contained in the imagingbook library has been updated and enhanced (see package imagingbook.pub.color.quantize):

• KMeansClusteringQuantizer: new quantizer using k-means clustering,
• KMeansClusteringQuantizerApache: new quantizer using the Apache Commons Math k-means clustering implementation,
• MedianCutQuantizer: revised version,
• OctreeQuantizer: re-implemented and operational now.

All the above are subclasses of ColorQuantizer, which is now an abstract class (used to be an interface) and defines various common methods.

Demo ImageJ plugins are provided for the above classes in the imagingbook-plugins-all repository.