README.md

Plantar Pressure Reconstruction Using Compressive Sensing

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

Contents

• 1. Requirement
• 2. Dataset
• 3. Training and Evaluation
• Citation
• License

1. Requirement

• The codes are written with MATLAB R2016b.
• The mfiles KSVD.m, KSVD_NN.m, OMP.m, OMPerr.m, NN_BP.m, and my_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.

2. Dataset

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.

3. Training and Evaluation

• Run learningtest.m and set dict.learn flag to learn dictionary, and get evaluation plots for each specified subject and foot. You may set dict.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 set dict.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.

Citation

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}
}

License

• This code is for non-commercial purpose only. For other uses please contact ACLab of NEU.
• No maintenance service