论文部分内容阅读
激光成像雷达能成清晰的目标三维距离像和一维强度像,可提高目标识别率,因而成为国际上的研究热点。当大视场高分辨率激光成像雷达垂直探测目标时,视场内目标增多,要求目标识别算法既能同时检测多目标,又要具有平面内旋转不变性。为了满足上述要求,提出将具有平面内旋转不变性的CHF-MACH相关滤波器和支持向量机(SVM)相结合,组成一种新的目标识别系统,其中相关滤波器能同时检测定位多个感兴趣目标,再用SVM分别对图像内的已定位的目标进行识别。以仿真激光成像雷达图像为实验数据,分别对4类目标进行识别。实验结果表明,CHF-MACH滤波器对本类目标有较好的检测率,对非本类目标有一定的抑制作用;SVM能以较高的精度分类已检测目标。所以,该方法能有效地对大视场内多目标进行识别,适用于激光成像雷达。
Laser imaging radar can be a clear target three-dimensional distance image and one-dimensional intensity images, can improve the target recognition rate, which has become an international hot spot. When large-field high-resolution laser imaging radar targets are vertically detected, the number of objects in the field of view increases, which requires that the target recognition algorithm not only detects multiple targets simultaneously but also has in-plane rotation invariance. In order to meet the above requirements, a new target recognition system is proposed by combining CHF-MACH filter and SVM with in-plane rotation invariance. The correlation filter can detect and sense multiple senses at the same time Interest targets, and then use SVM to identify the target within the image has been targeted. Taking the simulated laser imaging radar image as the experimental data, the four kinds of targets are respectively identified. The experimental results show that CHF-MACH filter has good detection rate for this class of target, which has certain inhibitory effect on non-target. SVM can classify the target under test with high accuracy. Therefore, this method can effectively identify multiple targets in a large field of view and is suitable for laser imaging radar.