Laser vibrometry exploitation for vehicle identification

Effects of fundamental frequency normalization on vibration-based vehicle classification

“…A fair comparison of classification performances was desired so feature selection was excluded. Lastly, the calculated features were used to identify the vehicles with four classifiers from [2] and [5]. The four classifiers included Naïve Bayes [17] , Support Vector Machines (SVM) [18,19] , the C4.5 decision tree [20] , and k-nearest neighbors (KNN) with k=3 [2,21] in WEKA.…”

“…These features were taken from speech processing, seismic signal processing, and structural analysis. The eleven features include zero crossings, complexity, root mean square (RMS), linear prediction coefficients (LPC), dominant frequencies, Mel-frequency cepstral coefficients (MFCC), spectral flux, number of peaks, spectral rolloff, spectral centroids, and spectral ratios [2] . All of the features except LPC were utilized in this research; LPC was excluded because the method of calculation in [2] could not be verified.…”

“…Feature extraction in MATLAB is completed using the preprocessed data. The features used in the end-to-end process were selected from [2]. These features were taken from speech processing, seismic signal processing, and structural analysis.…”

“…Linear prediction coefficients (LPC) from directional microphone data were used to train time delay neural networks (TDNN) with 95-100% accuracy for a small set of traffic patterns and vehicle speeds [1] . Features drawn from speech processing, seismic signal processing, and structural analysis were utilized in [2] to identify stationary vehicles with classification rates between 70% and 100% dependent on the classifier. Recently, a paper was published utilizing the dominant frequency, or firing frequency, and the related half order frequencies to identify the fuel type and piston count of stationary vibration signatures collected with accelerometers [3] .…”

“…This paper investigates the effects of fundamental frequency normalization on the end-to-end classification process [1] . Using the fundamental frequency, assumed to be the engine's firing frequency, has previously been used successfully to classify vehicles [2,3] . The fundamental frequency attempts to remove the vibration variations due to the engine's revolution per minute (rpm) changes.…”