Skip to content
No description, website, or topics provided.
MATLAB
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
CA_mean.mat
ClassifyOnNN.m
LICENSE
NILBP_Image.m
NewRDLBP_Image.m
RDLBP_Image_SmallestRadiusOnly.m
README.md
README_Yuting.docx
ReadMe.txt
SSELBP_Feature_Extraction.m
SSELBP_KTHTIPS.m
cirInterpSingleRadius.m
cirInterpSingleRadiusNew.m
dataset.txt
distMATChiSquare.m
get_img_path.m
get_mapping_info.m
get_outexProbSet.m
get_subPath.m
getmapping.m
license.txt
matlab.mat
samp_prepro.m

README.md

SSELBP_ICASSP2017

Publication: Y. Hu, Z. Long, G. AlRegib, “Scale Selective Extended Local Binary Pattern for Texture Classification,” IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP2017), pp. 1413-1417, 2017.

Bib: @inproceedings{hu2017scale, title={Scale selective extended local binary pattern for texture classification}, author={Hu, Yuting and Long, Zhiling and AlRegib, Ghassan}, booktitle={Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on}, pages={1413--1417}, year={2017}, organization={IEEE} }

Objective: The codes are used to verify the performance of the scale selective extended local binary pattern (SSELBP) on the classification accuracy on the database of KTH-TIPS (under KTH_TIPS directory).

Key functions and variables:

----- SSELBP_Feature_Extraction.m Test the performance of SSELBP on texture classification.

  1. Sampling scheme a. Variable “scheme” specifies a set of radii for sampling locations; b. Variable “lbpPointsSet” denotes a set of the number of sampling points on each circle; c. Variable “mapping” defines a mapping strategy and we use “riu2” here.
  2. Features and labels a. Variable “mrelbp_tests”: each row contains the histogram feature and label for each test image; the last column represents class labels; b. Variable “mrelbp_trains”: each row contains the histogram feature and label for each training image; the last column represents class labels; c. Variable “fd_scale” represents the feature dimension for each scale.

----- SSELBP_KTHTIPS.m Extract SSELBP features for each predefined (P, R) at each scale in the scale space.

----- distMATChiSquare.m Calculate a distance matrix based on the chi-square distance.

  1. Distance matrix denoted by variable “DM”.

----- ClassifyOnNN.m Record the classification accuracy for each trial using nearest neighbor classifier.

  1. Variable “numSamples_class” denotes the number of samples for each class;
  2. Variable “numtrain_class” stands for the number of training samples for each class;
  3. Variable “numClass” defines the number of class;
  4. Variable “train_idx” and “test_idx” define the indices for random partitions between training samples and testing samples
  5. Variable “accuracySSELBP” stores the classification accuracy for 100 partition trails;
  6. Variables “CA_mean” means the average classification for each testing scheme.
You can’t perform that action at this time.