Lucene Image REtrieval
The LIRE (Lucene Image REtrieval) library provides a simple way to create a Lucene index of image features for content-based image retrieval (CBIR), which allows searching for similar images. The used features are taken from the MPEG-7 standard: ScalableColor, ColorLayout, and EdgeHistogram. Furthermore, methods for searching the index are provided.
| Tags | Software Development Libraries Java Libraries Scientific/Engineering Image Recognition multimedia Graphics |
|---|---|
| Licenses | GPL |
| Operating Systems | OS Independent |
| Implementation | Java |
Recent releases
- 23 Oct 2008 10:25
Changes: This is a major release. It fixes a lot of bugs and introduces several new features including a new descriptor, a simplified way to use descriptors by introducing new generic searchers and indexers, as well as a generalized interface for image descriptors. There are also several improvements in indexing and search speed. Furthermore, retrieval performance was optimized based on the Wang 1000 data set. All new features have also found their way into LireDemo.
- 09 Jun 2008 17:56
Changes: This release contains three additional features: Tamura, CEDD, and a color histogram implementation. CEDD and Tamura provide additional improvements. A FastMap implementation is included for optimization of the indexing process in a later release. Some bugs were fixed in the MPEG-7 EdgeHistogram descriptor provided in the cbir-library jar file and in color-only search. Note that due to the increased number of features, the extensive document builder needs significantly more time for extraction than in the last release.
- 10 Jul 2007 04:09
Changes: An issue where the scalable color descriptor (color histogram) was not compliant to the MPEG-7 standard was fixed. The color only search was changed to use the color layout descriptor and a bug in the edge histogram descriptor was hunted down. The LireDemo GUI application has also been updated: A new function for creating image mosaics has been introduced and the indexing of digital photos is now faster than ever as only the EXIF thumbnails - if available - are used instead of the whole image.
- 26 Feb 2007 14:20
Changes: This release brings a new descriptor, which is kind of "more advanced version of a color histogram". The so called "color correlogram" is based on the probability to find pixels of certain colors in certain neighborhoods. In general, it is a new way to retrieve photos with LIRe based on color and color distribution, which might be very interesting for applications heavily depending on color. This release also contains the LIRe Demo, a desktop GUI application showing indexing and searching images using different descriptors.
- 14 Dec 2006 06:51
Changes: Since Lire already supports Color histograms (with the MPEG-7 ScalableColor descriptor), functions for searching for colors have been integrated by by adding a searcher for color only search operations, a document builder restricted to color, and a document factory for fast and efficient creation of documents describing images with one color only.
- All comments
Recent commentsRe: Linear Search Growth in relation to the number of Images
> I am trying to search on about 400,000
> images and it is getting slow. I am
> interested to know if you have thought
> about any ways to make the search
> nonlinear in growth?
>
> I haven't taken the time to understand
> the CBIR index string yet so I am not
> sure how it can be indexed. Hopefully
> there is a better solution than reading
> all the indexes for each search though.
Yeah I know this ... the problem is of course the linear search, although I thought it would be getting slow at around 1,000,000 images.
An index taking care of this would be in the common case a multidimensional access method. I already thought of implementing such a thing, but ... well ... time is the crucial point. Suggestions for Java code for R* trees and the likes in Lucene are very welcome, contributions also :-)
Linear Search Growth in relation to the number of Images
I am trying to search on about 400,000 images and it is getting slow. I am interested to know if you have thought about any ways to make the search nonlinear in growth?
I haven't taken the time to understand the CBIR index string yet so I am not sure how it can be indexed. Hopefully there is a better solution than reading all the indexes for each search though.