"""

Preprocess IMU data - filter and discretization

"""

import numpy as np

from config import *

from sklearn.cluster import KMeans

import os

from util import save_obj, load_obj

def parse_label(name):

for i in range(6):

if name.startswith(gesture_names[i]):

return i

def quantization_predict(*args):

return kmeans.predict(imu_measurements)

def quantization_train(imu_measurements, k):

return labels

if __name__=="__main__":

file_names = [p for p in os.listdir(file_path) if p.endswith('.txt')]

X = np.zeros([0, 6])

T = np.zeros([0])

ranges = [0]

print("loading data...")

for file_name in file_names:

# load data

data = np.loadtxt(file_path+file_name)

ts = data[:, 0]

imu = data[:, 1:7]

X = np.concatenate((X, imu), axis=0)

T = np.concatenate((T, ts))

ranges.append(ranges[-1]+np.shape(data)[0])

# quantize the measurement

print("clustering the measurement...")

labels = quantization_train(X, M)

print("save to new files...")

if not os.path.exists(quant_file_path):

os.makedirs(quant_file_path)

for i, file_name in enumerate(file_names):

# split data

ts = T[ranges[i]:ranges[i+1]]

q_imu = labels[ranges[i]:ranges[i+1]]

np.savez(quant_file_path+file_name[:-4], ts=ts, Z=q_imu)

# for new measurements:

# Z = quantization_predict(X)