一种基于改进地貌形状上下文的形状匹配方法

Shape Matching Method Based on Improved Aspect Shape Context

在基于地貌形状上下文的形状匹配方法中,计算地貌空间测地距离消耗时间较高,对应形状特征提取过程的效率较低.针对这一问题,本文提出了一种基于地貌模糊形状上下文的快速形状匹配方法.在形状特征提取过程中,通过引入最短路径算法对轮廓采样点间的测地距离进行快速计算.在此基础上结合对数极坐标模糊直方图构造地貌模糊形状上下文,其能够更好地描述轮廓点分布情况进而有效提升形状描述符的表达能力.考虑到轮廓点集顺序已知,进一步引入动态规划分析不同地貌空间下形状片段间的对应关系,以获取准确的形状匹配结果.通过对不同的数据集进行实验仿真分析,验证了本文方法能够有效地提升运算效率并取得较好形状检索精度.

In shape matching method based on aspect shape context, it is time consuming to calculate the geodesic distances on the aspect spaces, and the process of shape feature extraction is inefficient. To solve this problem, this paper proposes a fast shape method based on aspect fuzzy shape context. During the process of shape feature extraction, the geodesic distances between sample points on the shape contour can be effectively obtained by using the shortest path algorithm, and log-polar fuzzy histogram is further introduced to construct aspect fuzzy shape context, and the description ability is improved with the sample point distributions represented precisely. With the orders of sample points, the dynamic programming method is employed to analyze the correspondence between shape segments in different aspect spaces, and the shape matching result can be obtained accurately. With the proposed method tested on different shape databases, the computation efficiency can be effectively improved and desirable shape retrieval results can be achieved.