基于边缘特征的异质图像目标配准方法

Object Detection Method Using Multi-modal Images Registration Based on Edge Features

目标检测作为计算机视觉的重要研究课题,在视频监控和智能导航方面有重要应用。图像配准技术是刚体目标检测方法之一。异质图像间目标的纹理差异极大,为此,对于目标的可见光俯视图像与机载红外摄像机观测异质图像,通过研究基于边缘特征的配准技术,实现对地面平面目标的配准检测。笔者首先通过建立机载运动摄像机空间模型,解决了红外观测图像的视角变换问题,将其换到与参考图像一致的视角,提出一种新颖的基于稳定区域轮廓合成的边缘提取方法,更好地控制异质图像的边缘提取效果,边缘特征匹配是在距离变换的基础上实现的,提出了一种基于稳定区域距离图像互相关的匹配度量准则。实验表明,算法能有效地实现异质图像间的目标配准检测。

Object detection,as an important research field of computer vision,is widely used for video surveillance,automotive navigation,etc.Image Registration is an important method for rigid object detection.In multi-modal images,the texture of objects is quite different.In this paper,the multi-modal image registration method based on edge features,between visible orthophoto map and airborne infrared images,is researched to realize the planar object detection in ground scene.Firstly,an airborne camera model is built to correct the viewpoint for infrared sensed image,making the shape of objects unified in two images.Then a novel method is proposed,which is combining the contours of stable regions to attain the image edge features,and the correlation coefficient between edge distance images of regions is taken as a new metrics for edge matching.Experiments show that our method is effective for the objects registration between multi-modal images.