基于Hausdorff距离的飞机定位方法

为了更快、更准确地识别出遥感图像中的飞机目标,用角点与Hausdorff距离相结合的方法来定位飞机目标.首先对图像进行Harris角点提取,由于传统的Harris角点提取方法对尺度比较敏感,所以采用Harris-Laplacian角点提取方法,由于在尺度空间的每层图像上计算Harris角点的计算量比较大,结合机场图像背景单一且飞机的灰度值比较高的特殊性,提高角点检测的速度和准确性;然后利用改进的Hausdorff距离即基于平均距离值的Hausdorff距离对两个特征点集进行匹配来定位飞机目标.该方法只需要一个模板就能对飞机目标进行定位.通过对机场图像的试验结果表明,该方法能很好地定位出飞机目标,具有较好的鲁棒性和实用性.

Airplane locating method based on Hausdorff distance

In order to rapidly and correctly recognize the airplane in remote sensing images, a method combining corners and Harsdorff distance to locate the airplane was proposed. First, the method extracted Harris comers in an image, since the method of traditional Harris comers' extraction was sensitive to scale, the method of Harris-Laplacian comers' extraction was adopted. But the computation of Harris comers' extraction in every scale is expensive, therefore, the particularity of airplane image that the background is single and the airplane' pixel is higher was considered, so the speed and accuracy of corners' extraction could be increased. These feature points were matched based on improved Hausdorff distance. The proposed method needs only one template to locate the airplane, and experimental results show that the method is robust and practical.