摘要:
Recognition and classification of wood floor surface texture is an important link in the production process of solid wood flooring. Classifing floor according to texture direction and distribution by machine vision technology is helpful to improve production efficiency and sorting quality, which is of great significance to the transformation and technical innovation of automatic detection and classification technology in solid wood flooring production. The texture structure of the floor blocks is detailed and complex, and the change is irregular, which has been bothering the academia all the time. Up to now, there is no uniform definition of texture. The new computer vision technology developed in recent years has become one of the hot research topics in image processing,in which,one-class SVM,as an unsuper-vised classification model which can be used for classification and recognition based on unlabeled data samples, is commonly used in text classification, abnormal point recognition and so on. In this study, OCSVM model is used to classify the texture features of straight and curved floor blocks, which can real-ize the fast and automatic recognition of floor texture with higher accuracy.%地板块表面纹理识别和分类是实木地板生产过程中的一个重要环节,利用机器视觉技术对地板块按纹理走向、分布等进行分类,有利于提高生产效率和分选质量,对实木板材生产中的自动检测分级技术的改造与技术创新都有重要意义.地板块纹理结构精细复杂,变化无规则,一直困扰着学术界,到目前为止对纹理未有一个统一的定义.近年发展起来的新型计算机视觉技术,成为当前图像处理热点研究课题之一,其中One-class SVM作为一种无监督的分类识别模型,能够根据未标注数据样本进行分类识别,常用在文本分类、异常点识别等方面.本研究应用OCSVM模型对直纹、弯纹地板块的纹理特征进行分类,能够实现地板块纹理的快速、自动识别,有较高的准确率.