基于双界线的Otsu阈值分割法及其快速算法

发布时间：2018-05-29 19:06

  本文选题：阈值分割 + 线性拟合　； 参考：《计算机应用研究》2016年12期

【摘要】：传统二维Otsu阈值分割法由于运算时间长、抗噪能力不足而在应用中受到限制。为了克服这些缺点,提出了一种基于双界线的二维Otsu阈值理论及其快速算法。在新的二维直方图中,两条平行于对角线的界线决定目标和背景区域的宽度,垂直于对角线的分割直线决定阈值大小。该算法运用Roberts算子和线性拟合法确定双界线,然后运用改进的Otsu法计算最佳阈值,最后对噪声区域进行后处理。实验结果表明,该算法不仅运算速度快,而且具备较好的分割质量和抗噪性能。同时,快速算法的引入,进一步降低了运算量,使得该算法具备更好的实时性。
[Abstract]:The traditional two-dimensional Otsu threshold segmentation method is limited in application because of its long operation time and insufficient anti-noise ability. In order to overcome these shortcomings, a 2-D Otsu threshold theory based on double boundaries and its fast algorithm are proposed. In the new two-dimensional histogram, the width of the target and the background region is determined by two boundaries parallel to the diagonal line, and the threshold value is determined by the segmentation line perpendicular to the diagonal line. In this algorithm, Roberts operator and linear fitting are used to determine the double boundary, then the improved Otsu method is used to calculate the optimal threshold. Finally, the noise region is post-processed. Experimental results show that the algorithm is not only fast, but also has good segmentation quality and anti-noise performance. At the same time, the introduction of fast algorithm further reduces the computation, making the algorithm has better real-time.
【作者单位】： 西南交通大学机械工程学院;
【基金】：国家自然科学基金资助项目(51275431)
【分类号】：TP391.41
，

本文编号：1952046