Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2014
Öğrenci: ÖZGE TOPUZ ALEMDAROĞLU
Eş Danışman: ÇAĞATAY CANDAN, SEYİT SENCER KOÇ
Özet:This study aims to experimentally investigate the feasibility of discriminating human motions with the help of micro Doppler features by using radar. In this work, the human walking simulator by V. Chen is examined and is modified according to requirements of the study. Then, the time-frequency distributions to obtain the spectrograms of human motions are examined and the Wigner Ville Distribution and Short Time Fourier Transform (STFT) are chosen for the application. After the simulation studies, experimental data is collected by using a ground surveillance radar. The first part of the experimental data consists of walking data with 7 realizations for the ranges of 150 meters and 1000 meters. The second part of the experimental data consists of 3 human subjects with 7 realizations for different human motions as walking, running, crawling, creeping and for different walking azimuth angles of 0°, 30°, 60°. After the collection of the experimental data, the sequence of signal processing steps, which are matched filtering, MTI filtering, windowing, FFT and CFAR are applied to the data to obtain the target range information. After that the micro Doppler feature extraction process is started. A high pass filter is designed and applied to the matched filtered matrix. After windowing on the high pass filtered output, the ranges with target are extracted. Then, STFT is applied to the range columns of the target to get the spectrogram. Some feature extraction methods are discussed and a set of features is chosen. Six features, which are torso frequency, bandwidth of the signal, offset of the signal, bandwidth without micro Dopplers, the standard deviation of the signal strength, the period of the arms or legs motions are extracted from the spectrograms of running, crawling, creeping and walking with azimuth angles of 0°, 30°, 60°. Lastly, a simple neural network based classifier is constructed. The classification performances of different human motions by neural network classification are examined.