首页|基于区域特征的图像配准方法

基于区域特征的图像配准方法

扫码查看
针对毁伤前与毁伤后图像因拍摄时间、角度不同和毁伤导致图像变化剧烈,造成图像无法配准的问题,提出一种基于区域特征的图像配准方法.该方法通过提取建筑物区域的形状特征和空间关系特征,构建建筑物区域的特征描述符,利用特征描述符的相似性完成图像配准.为了使特征描述符具有平移不变性、旋转不变性和尺度不变性,将描述区域与其他区域的中心距离、中心连线与长轴夹角的范围区间均匀分割,利用分割区间的索引构建二元组,通过计算满足二元组对应距离和夹角范围的建筑物区域与描述区域的面积比值作为特征描述符的元素.实验结果表明,与现有方法相比,该方法能完成毁伤前与毁伤后图像的配准,匹配率高,具有较好的鲁棒性.
An Image Registration Method Based on Regional Feature
Aiming at the problem that the images before and after damage cannot be registered due to the differ-ent shooting time,different shooting angles and drastic changes in the images caused by damage,an image regis-tration method based on regional features was proposed.The shape features and spatial relationship features of building area were extracted to construct the feature descriptors of building area,and the similarity of feature de-scriptors was used to complete image registration.In order to make the feature descriptor obtain translation in-variance,rotation invariance and scale invariance,the range interval of the center distance between the descrip-tion region and other regions,the angle between the center line and the long axis were evenly divided,and the index of the divided interval was used to construct the binary group.By calculating the area ratio between the building area and the description area satisfying the range of the corresponding distance and included angle of the binary group as the element of the feature descriptor.Images from the Russia-Ukraine war were used for regis-tration experiments.The experimental results showed that,compared with the existing methods,the proposed method could complete the registration of images before and after damage with high matching rate and good ro-bustness.

image registrationregional featuredamage image

唐超、宫久路、谌德荣、王泽鹏、刘绍荣

展开 >

北京理工大学,北京 100081

北京航宇天穹科技有限公司,北京 100043

毁伤图像 区域特征 图像配准

2024

探测与控制学报
中国兵工学会 西安机电信息研究所 机电工程与控制国家级重点实验室

探测与控制学报

CSTPCD北大核心
影响因子:0.267
ISSN:1008-1194
年,卷(期):2024.46(3)