Fall Detector Using Discrete Wavelet Decomposition And SVM Classifier

Abstract

This paper presents the design process and the results of a novel fall detector designed and constructed at the Faculty of Electronics, Military University of Technology. High sensitivity and low false alarm rates were achieved by using four independent sensors of varying physical quantities and sophisticated methods of signal processing and data mining. The manuscript discusses the study background, hardware development, alternative algorithms used for the sensor data processing and fusion for identification of the most efficient solution and the final results from testing the Android application on smartphone. The test was performed in four 6-h sessions (two sessions with female participants at the age of 28 years, one session with male participants aged 28 years and one involving a man at the age of 49 years) and showed correct detection of all 40 simulated falls with only three false alarms. Our results confirmed the sensitivity of the proposed algorithm to be 100% with a nominal false alarm rate (one false alarm per 8 h).

Decision Support System Proposal for Medical Evacuations in Military Operations

“…In order to define the chance of survival function, 7200 cases were generated that covered the entire four-dimensional parameter space quite well. These cases were used to train two non-linear SVM (support vector machine) networks [32][33][34] and the final result was a function that assigns the red class a range of 1-50%, the yellow class 51-99% and the green class 100%.…”

Fall Detection using Lifting Wavelet Transform and Support Vector Machine

“…Discrete wavelet transform (DWT) has been proposed for mobility monitoring, posture transition and activities classification in [18] using a single chest-mounted sensors. In [19], another frequency domain feature extraction method using short-time Fourier transform (STFT) was proposed to shorten the calculation time of DWT.…”

“…Unlike DFT in [4], LWT can be constructed from time series signal directly. Unlike DWT in [18], LWT does not require convolution, translation or dilation of traditional mother wavelets. Furthermore, LWT allows in place calculation, with no need for auxiliary memory.…”

Histogram of Oriented Gradients with Cell Average Brightness for Human Detection

“…In another test, instead of uniformly changing the brightness of the whole window, the brightness has been randomly changed at random regions of the image. First, each tested window has been divided into two parts by a diagonal line at a variable position and angle, then the brightness of a randomly chosen part has been changed according to (8) with the coefficients changed randomly for each window. The random brightness change has been also applied to the randomly selected rectangular areas of the window (with the random number of rectangles from 1 to 4).…”

“…The SVM is a well-known method of classification with a solid mathematical background, where the learning phase is reasonably short, but the classification stage requires a significant number of multiplications and additions. The SVM has been successfully applied to many different problems [7,8]. The boosting classifier consists of a cascade of "weak" classifiers, where early stages of the cascade reject most negative data.…”