A Wood Measuring System Based On Computer Vision

Wood measurement is an important part of wood production and processing.The cost and accuracy of the wood measurement process affect the whole wood industry and the protection of the forest resources to a great extent.In view of the problem of wood measurement,many countries in the world have designed a variety of automated wood measuring system,some of which are relatively mature.Because of the convenience and the cost,we still use manual method in most of the production environment in our country,so it is very important to design a simple and low cost wood measuring system.From the point of view of computer vision,this paper designs a kind of wood measuring system for measuring the cross section of wood through multiple shooting by a single camera.In this method,the object detection algorithm based on deep learning is used to identify the main area in the image,and the valid region is segmented by image processing method.Then,the spatial information of cross section is restored by using the knowledge of binocular vision.Finally,the cross section measurement of wood is completed by using the knowledge of spatial geometry.The main work and contributions of this paper are as follows:(1)According to the characteristics of pile sheets,this paper designs based on computer vision,non-contact wood measurement system,overcoming the defect of traditional measurement systems which can only measure logs,successfully solves the problems of measurement pile sheets,at the same time,makes the measurement process easier and the cost lower.(2)This paper introduces the more advanced methods of depth learning into the field of wood measurement,and deeply analyzes and discusses various object detection models based on depth learning.Finally,it is found that the SSD model(Single Shot MultiBox Detector)can not only achieve faster speed,but also have higher accuracy,and it is more perfect for wood cross section detection.The intervention of deep learning method makes the measurement work more intelligent.In the complex environment,it can still identify the subject to be tested.The robustness of the measuring system is greatly improved.