The invention relates to a vehicle classification method based on a single frequency continuous wave radar, which belongs to the field of intelligent vehicle detection. The method based on the time domain radar signal single frequency continuous wave radar as input, the time-frequency analysis of radar echo Doppler spectrum changes with time; by Hough transform Doppler spectrum image is mapped to the vehicle scattering center position parameters through image; feature extraction; KarhunenLoeve (K-L) screening and compressed samples, sample classification the Fisher criterion is used to obtain the vehicle type classification. In the case of multiple types, two types of classifiers are used to classify the 22 classifiers. Single frequency continuous wave radar has simple and reliable structure and low cost. The invention uses the single frequency continuous wave radar to realize the vehicle classification at the same time of vehicle detection and speed measurement, and the average classification accuracy rate is 94.8% when the vehicle is divided into the following categories in practical application.
【技术实现步骤摘要】
【技术保护点】
基于单频连续波雷达的车型分类方法,其特征在于,该方法步骤如下:步骤一,将时域雷达信号进行时频分析得到回波信号的多普勒频谱图t-f↓[d]图像空间:采用短时傅立叶变换STFT:STFT{x[]}≡X(m,ω)=*x[n ]w[n-m]e↑[-jωn]其中x[n]是离散时间信号,w[n]是窗函数,m是窗函数的滑动位置,ω是角频率;车辆散射中心产生的回波信号的多普勒频谱图t-f↓[d]图像空间为:***其中,x↓[v]为车辆散射 中心横坐标,h↓[v]为车辆散射中心纵坐标,x↓[0]为车辆入场位置,h↓[0]为雷达安装高度,f为雷达工作频率,c为光速,f↓[d]为车辆散射中心反射回波的多普勒频率,v为车辆相对速度;步骤二,利用哈夫变换将多普勒频谱图t-f↓[ d]图像空间转换为参数x↓[v]-h↓[v]空间,得到参数x↓[v]-h↓[v]空间中车辆散射强度的分布,参数空间中的曲线为:h↓[v](x↓[v])=h↓[0]-(x↓[v]+x↓[0]-vt)*;在实现哈夫变换的 ...
【技术特征摘要】
【专利技术属性】
技术研发人员:孟华东,张颢,王希勤,房建新,
申请(专利权)人:清华大学,
类型:发明
国别省市:11[中国|北京]
还没有人留言评论。发表了对其他浏览者有用的留言会获得科技券。