Local Intensity Order Pattern for Feature Description

IEEE International Conference on Computer Vision(ICCV 2011)

Zhenhua Wang, Bin Fan and Fuchao Wu

The extension of this ICCV paper has been accepted by TPAMI, please refer to the new project page!

  

Abstact

This paper presents a novel method for feature description based on intensity order. Specifically, a Local Intensity Order Pattern(LIOP) is proposed to encode the local ordinal information of each pixel and the overall ordinal information is used to divide the local patch into subregions which are used for accumulating the LIOPs respectively. Therefore, both local and overall intensity ordinal information of the local patch are captured by the proposed LIOP descriptor so as to make it a highly discriminative descriptor. It is shown that the proposed descriptor is not only invariant to monotonic intensity changes and image rotation but also robust to many other geometric and photometric transformations such as viewpoint change, image blur and JEPG compression. The proposed descriptor has been evaluated on the standard Oxford dataset and four additional image pairs with complex illumination changes. The experimental results show that the proposed descriptor obtains a significant improvement over the existing state-of-the-art descriptors.

Experiment Results


For more detailed results, please refer to the paper

Download

Paper: pdf
Poster: jpg
BitTex: bib

Test Data:
The Oxford 2D datasets can be downloaded here, and the complex illumination datasets can be downloaded here.

Code:
The affine covariant region detector used in our experiments can be downloaded from the Oxford University: Affine Covariant Region Detector.
The SIFT descriptor used in our experiments can also be downloaded from the Oxford University: Region Descriptor.
The newest LIOP descriptor and the evaluation code can be downloaed from here: LIOP_Evaluation.zip

News:

2016 Feb. We have made our source code public available on github, try it now! https://github.com/foelin/IntensityOrderFeature

2013 Jun. We have contributed our codes to the VLFeat open source library. The tutorial page can be found here, and the LIOP source codes can be found in the latest version(0.9.17).

Updates:

Name Version Comments

LIOP Binaries

Windows 32bit v1.2-20111224 Fix a bug in v1.1,and release the binaries for both Windows and Linux
Windows 64bit
Linux 64bit
LIOP Binaries for Windows 32-bit v1.1-20111201 Optimize the codes and speedup the descriptor construction about two times.
LIOP Binaries for Windows 32-bit v1.0-20111118 Orignal version used in the ICCV'11 paper

If you have any questions about this paper, please contact the author (zhwang dot me [at] gmail dot com).