Acoustic Noise Pattern Detection and Identification Method in Doppler System

Abstract

This paper discusses detection and identification of the moving vehicles based on prerecorded acoustic noise patterns. Object movement and object type differences would lead to the significant vehicle pattern changes. Doppler shift estimations of the measured signal spectral components will improve the recognition and classification performance. The proposed pattern identification methods could be used for the robust single sensor system or as a part of complex multisensor and learning solutions. The experimental results illustrate measured signal processing stages.

Hierarchical optical character recognition system design based on the Hopfield neural networks

“…Algorithm described in this paper represents a base for more sophisticated pattern recognition methods that can have multiple purpose in military applications. Pattern recognition algorithms play an important role in any kind of automated recognition method, such as target recognition in missile guidance, UAV image processing: recognition of visual landmarks, automated vehicle detection and classification by a typical noise pattern recognition and many others [11], [12] A good example is also the multiple-target tracking (MTT), an essential requirement for the surveillance systems employing one or more sensors, along with the computer subsystems. Primary sensors for the detection and target tracking are radars, as the active sensors, but they are increasingly supplemented with the passive sensors as infrared cameras.…”

Spectral features for audio based vehicle and engine classification

“…Still, our results can be improved, and we hope to get better results when more data are collected. Also, taking Doppler effect into account may further improve the results, see Berdnikova et al (2012), where data were compared with prerecorded sounds.…”