Compressed-sensing-assisted Target Recognition based on Wide-band Propagation Characteristics for Millimeter-wave Frequencies

In this paper, we made a study of target recognition for 39 GHz wide-band millimeter-wave frequencies. The propagation characteristics of the wide-band echo were chosen for feature extraction and target recognition. Measurements were conducted in outdoor environments with different types of targets by using a vector network analyzer (VNA). A Compressed sensing (CS) method was performed on the measured high resolution range profile (HRRP) data by using orthogonal matching pursuit (OMP) algorithm. Reconstructions of the target signal HRRPs were performed and the features of the target signal were successfully extracted. Then, the k-nearest neighbors (kNN) algorithm and the support vector machine (SVM) algorithm were used respectively to train the extracted features which are further applied for target recognition. The recognition performances of both algorithms were analyzed. Compared with traditional methods, the proposed method by using compressed sensing on radar HRRPs has significantly reduced the data storage in the stage of measuring and shows high accuracy in recognition.