| 摘要: |
| 针对利用单一特征进行地面轮式/履带式运动车辆目标分类时识别率低的问题，提出了基于证据理论的运动车辆多域特征融合识别方法。通过频域、时频域微动特征提取获得基本信任赋值，首先利用D S证据理论对频域特征结果进行融合得到其识别结果与识别信度，然后对频域与时频域识别结果进行决策级证据理论融合，得到运动点迹的目标类型，最后根据航迹判别规则完成轮式与履带式车辆的分类。通过实测数据试验验证，多域特征融合识别方法可以有效地利用地面动目标频域与时频域特征，进而提高了识别性能。 |
| 关键词: 目标识别 微多普勒特征 融合识别 证据理论 |
| DOI：DOI:10.3969/j.issn.1672-2337.2019.06.013 |
| 分类号:TN957.52 |
| 基金项目: |
|
| Study on Vehicle Micro Motion Feature Recognition Based on Evidence Theory |
|
ZHANG Huan，LIU Yujing，SUN Yongguang
|
|
AVIC Leihua Electronic Technology Research Institute，Wuxi 214063，China
|
| Abstract: |
| Aiming at the problem of low recognition rate when single feature is used to classify wheeled and tracked vehicles，a moving vehicle multi domain features fusion recognition method based on evidence theory is presented. The basic probability assignment function (BPAF) is gotten with the micro Doppler feature extraction. Firstly，the DS evidence theory is used to do feature level fusion recognition in frequency domain to get the result and confidence. Then，the decision level fusion based on DS in frequency domain and time frequency domain is carried out. Finally，according to the track recognition rule，the target type is recognized. The experimental results based on the measured data show that the method can efficiently utilize moving vehicle multi domain features and improve the recognition performance. |
| Key words: target recognition micro Doppler feature fusion and recognition evidence theory |
