This is the code for the following paper:
Farnoosh, A., Ostadabbas, S., & Nourani, M. (2017, November). Spatially-Continuous Plantar Pressure Reconstruction Using Compressive Sensing. In Machine Learning for Healthcare Conference (pp. 13-24).
Contact: Amirreza Farnoosh, Sarah Ostadabbas
- The codes are written with MATLAB R2016b.
- The mfiles
KSVD.m
,KSVD_NN.m
,OMP.m
,OMPerr.m
,NN_BP.m
, andmy_im2col.m
are taken from here with slight changes. These mfiles implement K-SVD algorithm proposed in: The K-SVD: An Algorithm for Designing of Overcomplete Dictionaries for Sparse Representation, written by M. Aharon, M. Elad, and A.M. Bruckstein.
The dataset comes from the original work of A knowledge-based modeling for plantar pressure image reconstruction and is placed in ./Dataset/
directory. It includes plantar pressure readings from 5 healthy subjects (right/left foot), and is augmented with the corresponding GMM centroids and variances from their method, as well as other information/preprocessing needed for our method.
-
Run
learningtest.m
and setdict.learn
flag to learn dictionary, and get evaluation plots for each specified subject and foot. You may setdict.learn = 0
if you wish to use pretrained dictionaries in./Dictionaries/
directory to get evaluation plots. -
Run
analyzeAll.m
to get aggregated evaluation results for all subjects and feet. Again setdict.learn = 0
if you want to use the pretrained dictionaries. -
Uncomment the corresponding lines (40-43) in
testallfunc.m
if you want to switch between sparse reconstruction (fpreconst
), interpolation (fpreconst_interp
), and GMM method (fpreconst_gmm
). -
Edit line 32 of
testallfunc.m
to specify the set of number of sensors (K) for which you want to get evaluation results. By default K is from 4 sensors (atleast) to 46. -
Set the corresponding figure flags to get the desired graphs.
If you find our work useful in your research please consider citing our paper:
@inproceedings{farnoosh2017spatially,
title={Spatially-Continuous Plantar Pressure Reconstruction Using Compressive Sensing},
author={Farnoosh, Amirreza and Ostadabbas, Sarah and Nourani, Mehrdad},
booktitle={Machine Learning for Healthcare Conference},
pages={13--24},
year={2017}
}
- This code is for non-commercial purpose only. For other uses please contact ACLab of NEU.
- No maintenance service