摘要
锯材的表面质量对木结构制品的质量起着至关重要的作用。锯材的表面质量主要受锯材表面缺陷(如节子、虫眼、裂纹等)类型、分布、数量等综合影响。文中针对基于浅层学习网络和深度学习算法的2类图像处理方法,对比分析其在锯材表面缺陷识别中的研究现状、存在的问题及发展趋势;同时,结合准确率、平均识别精度均值和图像识别时间等目标检测评价指标,对比分析多种图像处理算法在锯材表面缺陷识别中的性能;最后,对未来锯材表面缺陷识别方法的发展趋势进行了展望。
The surface quality of sawntimber plays a vital role in the quality of engineered wood products.The surface quality of sawntimber is mainly affected by the type,distribution and quantity of surface defects(including knots,insects,cracks,etc.).For the two types of image processing methods of shallow learning networks and deep learning algorithms,the paper compares and analyzes their research status,existing problems and development trends,and utilizes target detection evaluation indicators such as accuracy,mean average precision(mAP)and image detection time to conduct a comparative analysis of the performance of multiple image processing algorithms in the recognition of sawntimber surface defects.Finally,the paper prospects the future development direction of surface defect recognition methods for sawntimber.
作者
王勇
张伟
Wang Yong;Zhang Wei(Beijing Forestry Machinery Research Institute,National Forestry and Grassland Administration,Beijing 100029,China;Research Institute of Wood Industry,Chinese Academy of Forestry,Beijing 100091,China)
出处
《世界林业研究》
CSCD
北大核心
2022年第4期47-52,共6页
World Forestry Research
基金
中央级公益性科研院所基本科研业务费专项资金重点项目“木结构预制构件自动柔性加工技术与装备研发”(CAFYBB2018ZB011)
国家林业和草原局林业科学技术推广项目“木结构锯材应力分等装备制造技术应用与推广”([2019]35)
国家自然科学基金面上项目“结构用足尺锯材弹性模量快速评价体系与模型构建”(31670721)。
关键词
锯材
表面缺陷
识别方法
深度学习
图像处理
sawntimber
surface defect
recognition method
deep learning
image processing
