![]() ![]() The DC component suppression can be expressed asĪs mentioned earlier, our proposed method is composed of two parallel signal processing: one is to find the moment when the motion changes, and the other is to discriminate the type of motion. The process of applying the mean subtraction method to the spectrogram can be expressed in two consecutive steps. ![]() Among these methods, we applied the mean subtraction method, which has low computational amount and excellent clutter suppression performance. Various methods, such as a mean subtraction method, a range profile subtraction method, a linear trend subtraction method, can be applied to suppress the DC component and the radar clutter. Thus, we propose a method to remove the DC component and the static clutter that degrade human motion detection performance. In other words, a signal preprocessing step to remove those unnecessary signals is required before generating the input data used for motion identification. To extract only the signals corresponding to human motions, such unnecessary signals must be removed. The spectrogram described in Section 2.1 contains the DC component of the baseband signal and the static clutter. In our study, different CNN structures are determined according to the number of classes to be distinguished, and then classification performance is evaluated for each structure. For example, CNN models to classify hand gestures were designed, and CNN-based classifiers for classifying radar waveforms were also proposed. Recently, many studies have been conducted to classify targets by training radar data with the CNN. Second, to distinguish each motion, a convolutional neural network (CNN)-based classifier is trained with the cropped spectrogram. First, statistical characteristics appearing in the cropped spectrogram are used to find out the moment when the movement changes. Then, two major signal processes are performed in parallel in our proposed method. ![]() The extracted spectrogram is not used as it is, but the cropped spectrogram is used considering the area where the human exists. Therefore, we generate a spectrogram from the received radar signal and suppress the direct current (DC) component and the static clutter to extract only motion information from it. To identify the motion of a person, it is necessary to understand the change in radar signal characteristics over time. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |