Design Of Color Sorting System For Solid Wood Floor Based On Machine Vision

The color classification of solid wood floor based on machine vision has characteristics of improving the quality of solid wood floor,which also can drive the upgrading of the industry.At present,the production of solid wood floor has the difficulty of low classification efficiency,high labor cost and intensity due to manual classification.With the development of machine vision,its combination with industrial production become more and more closely,which makes industrial production intelligent and automatic.Therefore,the research on color classification system of solid wood floor has important significance and application value.This subject aims to solve the problem of color classification of solid wood floor combined with machine vision,so that the system can classify solid wood floor automatically.(1)First of all,this subject analyzes the present situation of the solid wood floor production.By combining with the actual user requirements,the existing problems in the current production was put forward.Based on this,the color classification of wood floor by machine vision was proposed,including designing the overall structure of the system,proposing the overall design method,and analyzing the different modules of the system.(2)According to the requirements of production capacity and the feature of machine vision product,the model of all the equipment were determined.As for the classification system,the lower control system based on STM32F407 was designed to realize the control of light source,linear scan camera,calibration mechanism and other equipment.Beside this,a solenoid valve driving unit with CPLD as the core was used to control solenoid valve by receiving the LVDS signal form the lower control system.(3)Then,the image processing algorithm by combining probability statistics and machine learning was proposed to classify the solod wood floor.In this part,a feature extracting method base on probability(random sampling feature extractor)was proposed to extract color features rapidly,which also can avoid the interference of the wood grain.Then,a fast classification neural network was built by extreme learning machine(ELM),and the color features optimized by genetic algorithm(GA)were used as the input of the ELM.At last,the experimental results was discussed to evaluate the performance of the proposed method.The classification accuracy of this method for solid wood floor is 97.78%,and the classification time is less than 1ms.(4)Finally,the upper machine software that used to do the human-computer interaction and the lower control system software were designed,which integrated image acquisition,image processing and classification mechanism.The upper machine software was based on OpenCV and QT with the characteristic of modularization.The systems functions such as equipment testing,system parameter setting and production were integrated in the system.The lower control system based on STM32 micro controller would communicate with the upper machine in real time through Ethernet,so that it can cooperate with the upper machine to complete the classification of the solid wood floor.The solid wood floor classification system based on machine vision can classify the solid wood floor correctly by experimental verification.The system complete the process requirements of solid wood floor,which has practical application value.