Expert System For Nondestructive Testing Of Solid Wood Timber Based On Image Processing

China's forestry industry occupies an important position in the development of national economy,driving the rapid development of furniture,decoration and other industries.With more and more people's demand for furniture and decoration personalized and high-quality,the detection of defects and texture characteristics of solid wood timbers is particularly important.In recent years,the development of computer technology,artificial intelligence and network technology has also promoted the recognition of defects and texture of solid wood timbers from the initial manual recognition to machine recognition.The introduct ion of artificial intelligence has great significance for the transformation and upgrading of wood processing enterprises and the development of intelligent manufacturing.Based on this,this paper had studied the defect and texture detection and classification of solid wood timbers,designed a non-destructive detection expert system of solid wood timbers based on image processing,and applies image processing and in-depth learning technology to help wood proce ssing enterprises to quickly and efficiently grade wood quality.The main research contents were as follows:(1)Through investigation and research,the related requirements of wood processing enterprises were analyzed,and the corresponding design require ments were put forward.The overall framework of solid wood timbers non-destructive testing expert system was designed.According to the causes and characteristics of defects and texture of solid wood timbers,the grades of solid wood timbers were classified,and the knowledge base of solid wood timbers grades division was established.(2)The reasoning mechanism of defect classification of solid wood timbers was established.The defect contour of solid wood timbers was extracted by using the excellent feature extraction ability of convolution neural network.Combining with the classification ability of limit learning machine,the corresponding algorithm was adopted to optimize the classification.A classifier of limit learning machine based on depth structure was designed to extract and classify the defect of real wood.The performance of the algorithm was analyzed and compared.(3)The reasoning mechanism of wood texture type discrimination was established.The local binary mode was used to image processing of wood samples,the self-learning depth confidence network was constructed,the texture feature discrimination was realized,and the performance was verified by comparing with other algorithms.(4)By applying Visual Studio 2017 and SQ L Server 2012 software tools and TensorFlow 1.5.0 in-depth learning framework,an expert system for nondestructive testing was developed.The function modules and operation of the system were introduced.The image processing technology was integrated into the expert system,which had good human-computer interaction.