摘要
实木地板具有良好的美学性能,是室内装饰的优良材料。为了满足特定的室内装饰要求的艺术效果,实木地板的颜色需要协调,因此在生产过程中需要根据实木地板的颜色进行分类,以满足客户的个性化需求。然而,传统人工识别分选的方法受劳动强度、工作效率等因素制约,难以满足产业的发展需求。本试验基于机器视觉技术,对实木地板颜色的图像采集装置与识别分类算法进行了设计。首先,利用彩色CCD相机对3个颜色等级的实木地板进行图像采集,共获得了150幅木材图像,其中80%用来训练、10%用来测试、10%用来验证。然后,利用极限学习机ELM进行颜色分类,并比较不同优化算法的优化效果,如灰狼优化算法(GWO)、遗传优化算法(GA)、粒子群优化算法(PSO)。结果表明,利用灰狼优化的极限学习机(GWO-ELM)对3个颜色等级实木地板图像的识别准确率最高,分别为88.6%,86%和100%。考虑到实木地板的单块完整性,算法的分类结果能有效满足实际生产需求,为国内实木家居企业的智能化发展提供了一套行之有效的解决方法。
Solid wood flooring has good esthetic properties and is an excellent material for interior decoration,which is ushered in a rapid and substantial increase in consumer demands with the economic growth.To meet specific interior decoration requirements in an artistic view,the color of solid wood flooring needs to be coordinated.Thus,the color of the solid wood flooring needs to be sorted to meet the individual needs of customers during the production process.However,the traditional manual identification and sorting method is restricted by labor intensity,work efficiency and other factors,which is difficult to meet the development needs of the industrial production.Therefore,this experiment designed the image acquisition device and the recognition and classification algorithms of the color images of solid wood flooring based on machine vision technology.First of all,the color charged-coupled device(CCD)camera was used to collect solid wood floors of three color grades and obtained 150 wood images totally,80%of which were used as training set,10%as testing set and 20%as validation set.Secondly,the extreme learning machine(ELM)was used to perform color classification and the optimization effects were compared by different optimization algorithms,such as Gray Wolf Optimization Algorithm(GWO),Genetic Optimization Algorithm(GA),and Particle Swarm Optimization Algorithm(PSO).After the experiment,the recognition accuracies of the three-color grades of solid wood flooring images using the extreme learning machine optimized by the gray wolf optimization algorithm(GWO-ELM)were 88.6%,86%and 100%,respectively.Compared to the extreme learning machine optimized by the genetic optimization algorithm(GA-ELM)and the extreme learning machine optimized by particle swarm optimization algorithm(PSO-ELM),the GWO-ELM showed obvious advantages.Taking the integrity of the single solid wood flooring into account,the classification results of the algorithm can meet the needs of the actual production effectively and provide a set of solutions for the intelligent development of domestic solid wood furnishing enterprises.
作者
王锦亚
李振业
倪超
WANG Jinya;LI Zhenye;NI Chao(College of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing210037,China)
出处
《林业工程学报》
CSCD
北大核心
2021年第5期135-139,共5页
Journal of Forestry Engineering
基金
江苏省重点研发计划(产业前瞻与关键核心技术)项目(BE2019112)。
关键词
实木地板色差
机器视觉技术
分色识别
极限学习机
灰狼优化算法
color difference of solid wood flooring
machine vision
color recognition
extreme learning machine(ELM)
gray wolf optimization algorithm(GWO)
作者简介
王锦亚,男,研究方向为图像处理。;通信作者:倪超,男,教授。E-mail:chaoni@njfu.edu.cn。
