基于改进均值标准差曲线描述子的反射对称轴检测

发布时间：2018-06-16 12:07

本文选题：均值标准差曲线描述子 + 曲线匹配　；参考：《电子学报》2017年07期

【摘要】：针对反射对称轴难以检测的问题,该文提出一种基于改进均值标准差曲线描述子(Mean-Standard deviation Curve Descriptor,MSCD)的反射对称轴检测算法.该算法首先对MSCD曲线描述子进行改进,使其具有镜像反射不变性并实现对称曲线对检测;然后采用距离约束并使用Hough变换获取图像的局部对称轴;最后通过局部对称轴合并得到最终对称轴.实验结果表明,该算法可实现图像的单对称轴和多对称轴检测,在亮度变化、对比度变化、噪声污染、模糊以及形变情况下,均能够准确定位图像对称轴并具有较强鲁棒性.
[Abstract]:Aiming at the problem that the reflection symmetry axis is difficult to detect, this paper presents an algorithm for detecting the reflection symmetry axis based on the improved mean standard deviation curve descriptor (Mean-Standard deviation Curve). Firstly, the MSCD curve descriptor is improved to be mirror reflection invariant and the symmetric curve pair is detected, then the local symmetry axis of the image is obtained by distance constraint and Hough transform. Finally, the final symmetry axis is obtained by merging the local symmetry axis. The experimental results show that the algorithm can detect the single symmetry axis and the multiple symmetry axis of the image. The algorithm can accurately locate the symmetry axis of the image under the condition of brightness change, contrast change, noise pollution, blur and deformation, and has strong robustness.
【作者单位】：河南理工大学计算机科学与技术学院;西安电子科技大学计算机学院;
【基金】：国家自然科学基金(No.61272394,No.61472119,No.61572173,No.61472373,No.61401150) 河南理工大学创新型科研团队项目(No.T014-3);河南理工大学杰出青年基金(No.J2016-3) 河南省高校基本科研业务费(No.NSFRF1604)
【分类号】：TP391.41

【相似文献】