摘要
为了快速高效地采用图像的最大稳定极值区域(MSER)进行图像匹配,提出了一种针对归一化MSER区域的双层匹配策略。对于待匹配图像的MSER区域进行归一化之后,采用互相关性进行区域的粗匹配;再将候选匹配区划分为4×4个子区域,计算对应子区域的hausdorff距离,作为投票依据,根据投票多少从候选匹配对中选择最优匹配对作为最终的匹配结果。采用标准图像库及实拍红外图像进行了2组实验,实验结果表明,该方法能够较好地完成图像匹配任务。提出了一种针对MSER区域的双层匹配策略,采用标准图像库及红外实拍图像进行了匹配试验。实验结果表明,方法简单高效,匹配效果较好。
To quickly and efficiently conducting the image matching task using the Maximally Stable Extremal Regions（ MSER）,this paper presents a novaltwo-step matching strategy for the normalized MSER regions.After the MSER regions extracted from images to be matched are normalized,the coarse matching is performed by using cross correlation; the candidate regions are divided into 4 × 4subregions,the hausdorff distance of the corresponding sub-regions is calculated,and the results are used as the voting basis.Depending on the voting results,the best matching pairs are chosen from the candidates as the final matches. The two experiments are conducted by using standard image database and real infrared images respectively.The experimental results show that the proposed method can better complete the image matching task. This paper presents a two-step matching strategy for MSER regions,and the two matching experiments are conducted by using standard image database and infrared images respectively. The results show that this method is simple and efficient,and has better matching effect.
出处
《无线电工程》
2017年第10期68-72,共5页
Radio Engineering
