BayesianFilter.cpp 5.12 KB
Newer Older
mzed's avatar
mzed committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
/**
 * @file BayesianFilter.cpp
 * @author Jules Francoise jules.francoise@ircam.fr
 * @date 2013-12-24
 *
 * @brief Non-linear Bayesian filtering for EMG Enveloppe Extraction.
 * Based on Matlab code by Terence Sanger : kidsmove.org/bayesemgdemo.html
 * Reference:
 * - Sanger, T. (2007). Bayesian filtering of myoelectric signals. Journal of neurophysiology, 1839–1845.
 * 
 * @copyright
 * Copyright (C) 2013-2014 by IRCAM - Centre Pompidou.
 * All Rights Reserved.
 * 
 * License (BSD 3-clause)
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *
 * 1. Redistributions of source code must retain the above copyright notice,
 *    this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the copyright holder nor the names of its
 *    contributors may be used to endorse or promote products derived from
 *    this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

#include "BayesianFilter.h"
#include "filter_utilities.h"


#pragma mark -
#pragma mark Constructors
BayesianFilter::BayesianFilter()
{
    mvc.assign(channels, 1.);
    init();
}

BayesianFilter::~BayesianFilter()
{
    
}

void BayesianFilter::resize(std::size_t size)
{
    if (size > 0) {
        channels = size;
        init();
    }
}

std::size_t BayesianFilter::size() const
{
    return channels;
}

#pragma mark -
#pragma mark Main Algorithm
void BayesianFilter::init()
{
    mvc.resize(channels, 1.);
    output.assign(channels, 0.);
    prior.resize(channels);
    state.resize(channels);
    g.resize(channels);
    for (unsigned int i=0; i<channels; i++) {
        prior[i].resize(levels);
        state[i].resize(levels);
        g[i].resize(3);
        
        double val(1.);
        for (unsigned int t=0; t<levels; t++) {
            state[i][t] = val * mvc[i] / double(levels);
            val += 1;
            prior[i][t] = 1. / levels;
        }
        
        double diff = diffusion * diffusion / (samplerate * std::pow(mvc[i] / levels, 2));
        g[i][0] = diff / 2.;
        g[i][1] = 1. - diff - this->jump_rate;
        g[i][2] = diff / 2.;
    }
}

void BayesianFilter::update(vector<float> const& observation)
{
    if (observation.size() != this->channels) {
        resize(observation.size());
    }
    
    for (std::size_t i=0; i<channels; i++)
    {
        // -- 1. Propagate
        // -----------------------------------------
        
        vector<double> a(1, 1.);
        vector<double> oldPrior(prior[i].size());
        //        oldPrior.swap(prior[i]);
        copy(prior[i].begin(), prior[i].end(), oldPrior.begin());
        
        filtfilt(g[i], a, oldPrior, prior[i]);
        
        // set probability of a sudden jump
        for (unsigned int t=0; t<levels; t++) {
            prior[i][t] = prior[i][t] + jump_rate / mvc[i];
        }
        
        // -- 4. Calculate the posterior likelihood function
        // -----------------------------------------
        // calculate posterior density using Bayes rule
        vector<double> posterior(levels);
        double sum_posterior(0.);
        for (unsigned int t=0; t<levels; t++) {
            double x_2 = state[i][t] * state[i][t];
            posterior[t] = this->prior[i][t] * exp(- observation[i] * observation[i] / x_2) / x_2;
            sum_posterior += posterior[t];
        }
        
        // -- 5. Output the signal estimate output(x(t)) = argmax P(x,t);
        // -----------------------------------------
        // find the maximum of the posterior density
        unsigned int pp(0);
        double tmpMax(posterior[0]);
        for (unsigned int t=0; t<levels; t++) {
            if (posterior[t] > tmpMax) {
                tmpMax = posterior[t];
                pp = t;
            }
            posterior[t] /= sum_posterior;
        }
        
        // convert index of peak value to scaled EMG value
        output[i] = state[i][pp] / mvc[i];
        
        // -- 7. Repeat from step 2 > prior for next iteration is posterior from this iteration
        // -----------------------------------------
        copy(posterior.begin(), posterior.end(), prior[i].begin());
    }
}