摘要： 提出了一种基于三角剖分的形状检索改进算法。算法的基本思想是:先对图像进行边界跟踪和角点检测;然后寻找初始角点在边界跟踪中的对应点,并对找到对应点的角点按对应点在边界跟踪中的顺序进行排序;再对排序后的角点进行德洛内三角剖分,得到能表示目标真实形状的三角形序列;最后计算三角形序列的角度直方图作为形状特征进行相似性匹配。实验结果表明,该算法有较高的效率和检索 精度。

关键词: 形状检索, 三角剖分, 角点检测, 边界跟踪, 角度直方图

Abstract: An algorithm based on triangulation for shape-based image retrieval is proposed. In this algorithm, firstly, the edge of the original image is followed and the candidate corners in the original image are detected. Then the counterpoints of the candidate corners in the result of edge follow are found, and the boundary corners the counterpoints of which have been found are queued in the order of their counterpoints in edge follow. After these steps, the boundary corners are triangulated and triangles are obtained which can show the true shape of the original image. Finally, shape feature is obtained by calculating the angle histogram of the triangles. The experimental results show that the new algorithm has higher efficiency and retrieval precision.

中图分类号: 

• TP391.41