特征弦约束随机Hough变换在椭圆检测中的应用
发布时间：2018-08-14 11:05
【摘要】：为了提高复杂图像中椭圆/类椭圆的检测效率及精度,提出了一种基于特征弦约束的随机Hough变换(RHT)思想的椭圆检测改进方法,借助特征弦的几何约束及特征弦端点的法向约束,大幅度降低RHT的无效采样和累积次数。通过对边缘图像中像素点的有效分布进行分析,建立用于存储特征弦的端点信息的二维数组累加器,在边界点提取之前利用椭圆幂剔除虚假椭圆中心的干扰,不仅能够提高空间采样点的可靠性,同时降低无效采样点的累积概率。实验表明,该算法具有更高的运行速度和检测精度,同时对于形变较大,轮廓缺失严重及噪声具有较强的抵抗能力。
[Abstract]:In order to improve the efficiency and accuracy of ellipse/quasi-ellipse detection in complex images, an improved method of ellipse detection based on the idea of random Hough transform (RHT) with feature string constraints is proposed. By means of geometric constraints of feature string and normal constraints of feature string endpoints, the number of invalid sampling and accumulation of RHT is greatly reduced. By analyzing the effective distribution of pixels, a two-dimensional array accumulator is built to store the endpoint information of feature chords. Elliptic power is used to eliminate the interference of false ellipse center before extracting boundary points, which not only improves the reliability of spatial sampling points, but also reduces the cumulative probability of invalid sampling points. Higher operating speed and detection accuracy, and for large deformation, serious contour loss and strong resistance to noise.
【作者单位】： 长春理工大学光电工程学院;
【基金】：国家自然科学基金(61605016) 吉林省科技发展计划(20160520018JH)项目资助
【分类号】：TP391.41
[Abstract]:In order to improve the efficiency and accuracy of ellipse/quasi-ellipse detection in complex images, an improved method of ellipse detection based on the idea of random Hough transform (RHT) with feature string constraints is proposed. By means of geometric constraints of feature string and normal constraints of feature string endpoints, the number of invalid sampling and accumulation of RHT is greatly reduced. By analyzing the effective distribution of pixels, a two-dimensional array accumulator is built to store the endpoint information of feature chords. Elliptic power is used to eliminate the interference of false ellipse center before extracting boundary points, which not only improves the reliability of spatial sampling points, but also reduces the cumulative probability of invalid sampling points. Higher operating speed and detection accuracy, and for large deformation, serious contour loss and strong resistance to noise.
【作者单位】： 长春理工大学光电工程学院;
【基金】：国家自然科学基金(61605016) 吉林省科技发展计划(20160520018JH)项目资助
【分类号】：TP391.41
【参考文献】
相关期刊论文 前4条
1 杨险峰;蔡少虬;杨学友;;基于亚像素边缘的椭圆中心检测方法研究[J];光学仪器;2014年06期
2 王静;王海亮;向茂生;韦立登;刘忠胜;;基于非极大值抑制的圆目标亚像素中心定位[J];仪器仪表学报;2012年07期
3 段汝娇;赵伟;黄松岭;陈建业;;一种基于改进Hough变换的直线快速检测算法[J];仪器仪表学报;2010年12期
4 周小明;刘明;徐飞;;一种基于长轴和对偶性的椭圆检测新算法[J];光学技术;2007年05期
【共引文献】
相关期刊论文 前10条
1 蔡祥义;程天翼;吕p樣，
本文编号：2182651
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2182651.html
