| 摘 要: 木板材作为一种支撑社会发展的重要原材料，被广泛应用于建筑装修业和家具制造业，为了合理利用有限的木材资源，对其进行有效的检测至关重要。改进的Faster R-CNN(“两阶段”检测)算法对木板材活节、死节、孔洞、裂纹等四类缺陷检测的平均精度分别为99.84%、94.24%、91.28%、90.06%，平均精度均值为93.86%，并根据木板材缺陷类型对其进行等级划分。利用机器视觉引导技术，分拣机器人能够自动定位放置在传送带上的木板材，并依据木板材等级对其进行分拣作业。本文还基于C++语言、Qt框架搭建了用于支撑系统运行的软件平台。 |
| 关键词: 机器视觉；木板材分拣；机器人分拣；深度学习 |
|
中图分类号: TP302.7
文献标识码: A
|
|
| Research on Wood Panel Sorting System based on Machine Vision |
|
YAN Hengbing, ZHONG Liangwei
|
(School of Mechanical Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China)
8940212249@qq.com; zlvcad@126.com
|
| Abstract: Wood panel, as an important raw material to support social development, is widely used in building, decorating and furniture manufacturing industry. To make the most of limited wood resources, it is very important to detect wood panel effectively. The improved Faster R-CNN (Faster Region-based Convolutional Neural Network, a "two-stage" detection) algorithm has an average precision of 99.84%, 94.24%, 91.28%, and 90.06% for detecting wood panel live knots, dead knots, holes, and cracks, respectively. The mean average precision is 93.86%, and wood panels are classified by defect types. BY using machine vision guidance technology, sorting robots automatically locate the wood panels placed on the conveyor belt, and sort them according to the grade of the wood panels. A software platform for system operation is built based on C++ language and Qt framework. |
| Keywords: machine vision; wood panel sorting; robot sorting; deep learning |
