A Smartphone Step Counter Using IMU and Magnetometer for Navigation and Health Monitoring Applications

. 2017 Nov 8;17(11):2573. doi: 10.3390/s17112573

Algorithm 7. Step detection framework

Initiate adaptive filter frequencies and coefficients

Repeat for each sample:

------- peak extraction and update section -------

If in motion

Filter measurements

S ← Detect candidate

If S = 1

S_{k - 1}

= −1 && time_since_valley >

t_{a / r}

Valid peak detected

Elseif

S_{k - 1}

= S

Δ t = {t_{S}}_{K} - {t_{S}}_{K - 1}

{a_{p e a k}}_{k}

= peak update (

t h_{u}

S_{k - 1}

S_{k}

)

Δ t ≅

step duration && significant motion detected

Account for valley miss

Endif

Elseif S = −1

S_{k - 1}

= 1 && time_since_peak >

t_{a / r}

Valid valley detected

Elseif

S_{k - 1}

= S

Δ t = {t_{S}}_{K} - {t_{S}}_{K - 1}

{a_{v a l l e y}}_{k}

= peak update (

t h_{u}

S_{k - 1}

S_{k}

)

Δ t ≅

step duration && significant motion detected

Account for peak miss

Endif

------- peak validation from gyroscope and magnetometer -------

dominant axis extraction(ω(n), m(n))
If dr > 0

S_{ω}, S_{m}

← Detect candidate(ω(n), m(n))

S_{ω}

= 1 ||

S_{ω}

= −1

peak matching(

{t_{a}}_{p e a k}

{t_{ω}}_{p e a k}

{t_{m}}_{p e a k}

)

validate matching peaks

Endif

------- step validation -------

step validation (

{A_{s}}_{k}

)

adaptive filter frequency selector (LPD)