基于向量内积的新型骨架提取方法

摘要

A new method for skeleton extraction based on vector inner product was presented in order to improve the accuracy and connectivity.Euclidean distance transform is used to determine the nea-rest edge element for each pixel in a binary image.A vector from each pixel that stops at the nearest edge element is defined as edge vector.By comparing the inner product of edge vector within 8-neigh-borhood of the pixel,the number of significant changes in direction can be determined,and the candi-date skeleton points can be selected according to it.Finally,a complete skeleton was generated by ex-tending process based on regression analysis.Experimental results show that the algorithm can guarantee connectivity and integrity of the skeleton,and the average accuracy rate of location reached 92.27%.It also has advantages in reflecting the topological structures of objects and overcoming boundary disturbance.So,it is an effective skeleton extraction algorithm.%为了提高骨架提取的准确性和连通性，提出了一种基于向量内积的新型骨架提取方法。对二值图像进行欧氏距离变换，获得了由内部像素点指向边界点的边界向量，通过比较内部像素点8-邻域范围内对应边界向量内积值符号在4个方向上的变化情况确定了边界向量方向发生重大变化的次数，并据此选取候选骨架点；采用基于回归分析的方法确定延伸方向，并完成连接操作生成完整的骨架线。实验结果表明，该算法能够保证骨架的连通性和完整性，且骨架定位准确，平均正确率达到92．27％，同时可以克服边界扰动，是一种有效的骨架提取算法。