改进Canny算子对乐器板材的纹理分析  

Improved Canny Operator in Texture Analysis of Musical Instrument's Panels

作　　者：陈晓丹 于鸣[1] 黄英来[1] 高文韬 张定立 

机构地区：[1]东北林业大学信息与计算机工程学院,黑龙江哈尔滨150040

出　　处：《哈尔滨理工大学学报》2017年第4期46-50,57,共6页Journal of Harbin University of Science and Technology

基　　金：黑龙江省自然科学基金(C201347);大学生创新训练项目(201510225167)

摘　　要：针对人工肉眼观测板材的纹理走向、纹理断裂情况和纹理倾斜程度等带来的误差,提出了一种基于OpenCv和改进的Canny算子对板材图像进行纹理识别的方法。依据对Canny算子的分析与研究,针对Canny算子需要人工设置阈值的缺点,利用Otsu算法对其进行改进,实现阈值的自适应获取。实验结果表明,改进后的Canny算子比原始的Canny算子边缘检测效果更好,减少了边缘断裂和假边缘的情况,解决了去噪能力和保留边缘细节能力二者之间难以统一的矛盾,将其应用到乐器板材的纹理识别中,能够提高选材的准确率和效率。Due to the error cased byobservation of wood texturalfeatures including texture trend, texture fracture and texture gradient with eyes, a plate image texture recognition method which is based on Open Source Computer Vision Library and improved Canny operator is proposed. Because of the shortcomings of Canny operator that need to set the thresholds manually, It is successful in getting the threshold adaptively to use Otsu algorithm to improve Canny operator. The experiment result indicates that improved Canny algorithm which reduces the edge fracture and false edgeand resolves the conflict between denoising and keeping edge details is better than the original Canny in edge detection. The accuracy and efficiency of selection are improved by applying the improved Canny algorithm into wood texture recognition of national musical instruments.

关 键 词：CANNY算子 边缘检测 纹理识别 OTSU算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]