【作者】 马强； 高岚； 胡有为； 李小泉；

【Author】 Ma Qiang, Gao Lan, Hu Youwei, Li Xiaoquan(Energy and Power Engineering College of Wuhan University of Technology, Wuhan 430063)

【机构】 武汉理工大学能源与动力工程学院；

【摘要】 图像匹配的关键是时效性和准确性,同时也是判定算法优秀与否的重要标准。传统的基于灰度值的点与点之间做相关运算的算法虽然直接,精确度也较高,但是由于其搜索过程要遍历整幅图片,特别当图片为数据巨大的航空图片时,其花费的时间是巨大的,这不符合时效性的要求。Hausdorff距离是测量两个点集之间匹配程度的一种度量,在计算时不需要建立两个点集中点与点之间的对应关系,只需求得两幅图中相接近点的个数。采用部分Hausdorff距离定义,对目标有噪声影响的情况有较好的处理效果。采用Canny算子对图片提取边缘,得到图像中目标的边缘信息。仅利用边缘信息分析计算,而避免采用整幅图片的数据,大大缩短了计算时间。通过对多幅光学图片进行的检测,表明基于Hausdorff距离的综合匹配方法是有效的。更多还原

【Abstract】 The key points of target recognition are efficiency and accuracy, and also standards of judging whether a process is good or bad. Traditional algorithm is simple, which is based on relativity between point and point, and has high accuracy. This method needs to go through the whole picture, especially using picture that taken on plain, so the time spending on searching is too long and not fit for efficiency. Hausdorff distend is a method which is used for measuring the similarity of tow point sets. When we use it for calculating, we need not to establish the correspondence between the tow point sets, and only to calculate the close points. We can get a better result using the method of partial Hausdorff distance when the targets were affected by noise. We can get the edge information of the picture using Canny operator. Using this information for calculating and analyzing, which avoid adapting the whole data of the picture, we can save much time. Detected with some optical pictures, it shows that the synthetical matching process based on Hausdorff distance is effective.更多还原