Development Of Tobacco Image Acquisition Platform And Analysis Of Experimental Parameters

With the rapid development of digital image technology,intelligent tobacco leaf classification has become the trend of tobacco industry in the future,and tobacco leaf image quality is the key factor that affects the intelligent tobacco leaf classification.Most of current researches focus on the drive control and classification model of tobacco classifying equipment.There is rare research on the relationship between image acquisition conditions and classification accuracy.The main research contents of this paper are as follows:(1)Design of tobacco image acquisition platform.To solve the problem of poor adaptability of existing tobacco image acquisition equipment to different batches of tobacco leaves,the factors that affect the quality of tobacco images such as lighting conditions,shooting distance and shooting Angle were analyzed.In order to adjust the influencing factors and improve the applicability of the image acquisition platform,the scheme design of the image acquisition platform was carried out and the construction and hardware control of the image acquisition platform was completed based on the grading requirements of tobacco leaves.To achieve the unified deployment of hardware equipment,the communication software was studied for the tobacco image acquisition platform.(2)Realization of target location algorithm of tobacco image acquisition platform.The industrial camera selected by the platform was calibrated,and the internal and external parameter matrix of the camera were obtained.The radial distortion of the camera was corrected by using the parameter matrix.According to the color characteristics of tobacco leaves,the image threshold target detection algorithm was selected,and the median filter was used to reduce the image noise.The position coordinates of tobacco samples were extracted by means of minimum rectangular frame and coordinate transformation.(3)Study on correlation between control parameters of image acquisition platform and image quality.By adjusting the parameters of the platform,the conditions of image acquisition are controlled.Firstly,the platform is used to collect the image of the standard color card,and the influence of each condition on the color difference of the color card image is analyzed.On this basis,according to the characteristics of different batches of tobacco leaves,the acquisition conditions of the image acquisition platform were adjusted appropriately.The tobacco leaf images collected by the platform were used for subsequent tobacco leaf grading,and the grading accuracy was compared with that of the original tobacco leaf grading equipment.It was found that the accuracy of tobacco leaf grading could be improved from 61% to about 78% by using the tobacco leaf images collected by the platform.To sum up,this thesis explored the conditions of tobacco image acquisition,analyzed the key characteristics of tobacco leaf affecting the grading,developed the platform,and studied the correlation between the parameters of the platform and the image information.Tobacco leaf images collected by this platform were used for subsequent grading to verify the applicability of this image collection platform.