Abstract: The features of the defects image of wood knot are analyzed, and processing vector image as a whole one to protect its information, an algorithm of wood knot defect segmentation is proposed, based on extension of C-V vector model and AOS scheme coupling with filling background. A new level set approach is improved for vector image segmentation which is proposed by Chan-Vese based on Mumford-Shah, and the algorithm can increase the speed of segmentation. The AOS scheme is adopted to improve the original model of difference scheme, which is unconditional stable difference scheme. Combined with the technique of background a new image is made which reduces the different characteristics between the target and background. The experiment results show that, this method not only makes segmentation of the dead knot, live knot, burrow and veneer multi-knot defect images better, but also provides a new idea and method to edge segmentation of wood defect.

Key words: wood defect, extended C-V vector model, Additive Operator Splitting（AOS） scheme, veneer knot defect segmentation, painting background

摘要： 分析了木材节子缺陷图像的特点，将彩色图像作为一个整体的图像进行处理，保护了彩色图像信息的特性，提出了一种基于AOS的扩展C-V矢量模型及背景填充耦合的木材节子缺陷彩色图像分割算法。对Chan-Vese提出的基于Mumford-Shah模型的水平集矢量图像分割模型进行了改进，使分割速度得到了提高；用AOS算法改进了原模型的差分格式，使得差分格式无条件稳定；结合背景填充技术，使得到的新图像缩减了目标与背景间的特征差别。实验结果表明该方法可以较好地实现对木材死节、活节和虫眼等缺陷的彩色图像分割，也可实现对单板多节子缺陷彩色图像的分割，为木材缺陷边缘检测提供一种行之有效的方法。

关键词: 木材缺陷, 扩展C-V彩色模型, 加性算子分裂（AOS）算法, 单板缺陷分割, 背景填充