Added code, added README.md

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright {yyyy} {name of copyright owner}
+   Copyright 2017 LucaCerina
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

README.md

@@ -0,0 +1,15 @@
+README.md rev. 09 May 2017 by Luca Cerina.
+Copyright (c) 2017 Luca Cerina.
+Distributed under the Apache 2.0 License in the accompanying file LICENSE.
+
+# Automatic Multiscale-based Peak Detection (AMPD)
+
+ampdLib implements automatic multiscale-based peak detection (AMPD) algorithm
+as in An Efficient Algorithm for Automatic Peak Detection in Noisy Periodic and
+Quasi-Periodic Signals, by Felix Scholkmann, Jens Boss and Martin Wolf,
+Algorithms 2012, 5, 588-603.
+
+### Python required dependencies
+- Python > 3
+- Numpy
+

ampdLib.py

@@ -0,0 +1,98 @@
+# -*- coding: utf-8 -*-
+__author__      = "Luca Cerina"
+__copyright__   = "Copyright 2017, Luca Cerina"
+__email__       = "[email protected]"
+
+
+import numpy as np
+
+# AMPD function
+def ampd(sigInput):
+"""Find the peaks in the signal with the AMPD algorithm
+
+	Parameters
+	----------
+	sigInput: ndarray
+		The 1D signal given as input to the algorithm
+
+    Returns
+    -------
+    pks: ndarray
+        The ordered array of peaks found in sigInput
+"""
+
+	sigInput = sigInput.reshape(len(sigInput), 1)
+
+	# Create preprocessing linear fit	
+	sigTime = np.arange(0, len(sigInput))
+
+	fitPoly = np.polyfit(sigTime, sigInput, 1)
+	sigFit = np.polyval(fitPoly, sigTime)
+
+	# Detrend
+	dtrSignal = sigInput - sigFit
+
+	N = len(dtrSignal)
+	L = int(np.ceil(N / 2.0)) - 1
+
+	# Generate random matrix
+	LSM = np.random.uniform(1.0, 2.0, size = (L,N)) # uniform + alpha = 1
+
+	# Local minima extraction
+	for k in np.arange(1, L):
+		for i in range(k+1, N-k):
+			if((sigInput[i-1, 0] > sigInput[i-k-1, 0]) 
+			& (sigInput[i-1, 0] > sigInput[i+k-1, 0])):
+				LSM[k-1, i-1] = 0
+
+	# Find minima				
+	G = np.sum(LSM, 1)
+	l = np.where(G == G.min())[0]
+
+	LSM = LSM[0:l, :]
+
+	S = np.std(LSM, 0)
+
+	pks = np.flatnonzero(S == 0)
+	return pks
+
+# Fast AMPD		
+def ampdFast(sigInput, order):
+"""A slightly faster version of AMPD which divides the signal in 'order' windows
+
+	Parameters
+	----------
+	sigInput: ndarray
+		The 1D signal given as input to the algorithm
+	order: int
+		The number of windows in which sigInput is divided
+
+    Returns
+    -------
+    pks: ndarray
+        The ordered array of peaks found in sigInput 
+"""
+
+	# Check if order is valid (perfectly separable)
+	if(len(sigInput)%order != 0):
+		print("AMPD: Invalid order, decreasing order")
+		while(len(sigInput)%order != 0):
+			order -= 1
+		print("AMPD: Using order " + str(order))
+
+	N = int(len(sigInput) / order / 2)
+
+	# Loop function calls
+	for i in range(0, len(sigInput)-N, N):
+		print("\t sector: " + str(i) + "|" + str((i+2*N-1)))
+		pksTemp = ampd(sigInput[i:(i+2*N-1)])
+		if(i == 0):
+			pks = pksTemp
+		else:
+			pks = np.concatenate((pks, pksTemp+i))
+
+	# Keep only unique values
+	pks = np.unique(pks)
+
+	return pks
+