摘要
根据木材缺陷样本图像特点,用直方图均衡化和灰度变换对其进行预处理,使缺陷目标和背景反差更大,突出缺陷目标图像;对活节(样本1)、死节(样本2)、虫眼(样本3)缺陷采用多组对比试验的方法,采用传统分水岭算法和改进分水岭算法分割样本缺陷图像,分析两种方法的差异。结果表明:改进分水岭算法,能迅速且较好分割木材缺陷图像,缺陷轮廓更为清晰完整,曲线更为平滑,弥补了传统分水岭算法过渡分割和欠分割的不足,达到较为理想的分割效果;改进分水岭算法分割效率比传统分水岭算法更高,分割时间更短,更准确;试验证明了改进分水岭算法对木材缺陷图像分割的可行性和可靠性。
According to the characteristics of the wood defect sample image,it is preprocessed by histogram equalization and gradation transformation so that the contrast between the defect target and the background is larger,and the defect target image is highlighted.The method of multiple comparison experiments is applied to the defects of the live knots(sample 1),dead knots(sample 2)and wormhole(sample 3).The traditional watershed algorithm and the improved watershed algorithm are used to segment the sample defect images,and the differences between the two methods are analyzed.The comparison of experimental results show that the improved watershed segmentation method can quickly and better segment the image of wood defects,and the defect contours are more clear and complete,and the curves are smoother;this also makes up for the insufficiency of the traditional watershed transition segmentation and under-segmentation,achieving comparatively the ideal segmentation effect.The improved segmentation algorithm has higher segmentation efficiency than the traditional method,and the segmentation time is shorter and more accurate.This experiment proved the feasibility and reliability of the improved watershed algorithm.
作者
王金聪
宋文龙
张彭涛
Wang Jincong;Song Wenlong;Zhang Pengtao(Northeast Forestry University,Harbin 150040,P.R.China)
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2018年第10期93-97,共5页
Journal of Northeast Forestry University
基金
中央高校基本科研业务费专项资金项目(2572018BF14)
关键词
木材缺陷
木材图像分割
分水岭分割
Wood defects
Image segmentation
Watershed segmentation
