| Following with the development of computer technology,the image processing technology has been widely applied to agricultural engineering.But there is little application in the field of tea,especially in tea processing and tea cultivation.In this paper, the method and the feasibility of computer vision image processing technology using in the field of tea was explored,in order to provide ideas for implementing computer intelligence and fine operation in the field of tea.The main ccontents and results of this research were as following:(1) A new method of fetching unbroken tea color with computer,named L Method, has been established.The accuracy of the method was verified,and the further investigation also has been fulfilled.The results showed that the L method can fetch the color of tea accurately and quickly.There is little effect on fetching the color of tea by L method using Different scanning resolution.When the scanned area is 11 cm×11 cm or more,the color parameters of tea scanning photo may fully stand for the tea actual color. The SE is:-0.16≤L≤0.16,-0.13≤a≤0.13,b=0.There was obvious distinction between the color of unbroken tea and crush tea,so the later can't represent the tea actual color. Using computer to fetch color parameters is better than chroma meter.(2) A new method of image segmentation,that aimed at tea leaves and tea,has been established.Identified the best background color of tea leaves for image segmentation is blue,and the best tolerance is 150;the best background color of tea for image segmentation is red,and the best tolerance is 120.(3) A new method of quantizing the color of fresh tea leaves with computer has been established,and its exploratory development were studied initially.The results showed that it is reliable to quantize the color of fresh tea leaves with computer.The color parameters of R,G,B,L can identify the different tender degree of flesh tea leaves,the difference was significant at 0.01 level(R:F=64.26**;G:F=89.25**;B: F=46.91**;L:F=90.97**).The standard deviation of brigtness channel can reflect the uniform degrees and the freshness of tea leaves well.The color parameters of R,G, specially R,shows regularity variation in 20 hours after the fresh tea leaves have been picked.The color parameters of G,L,a,b,specially G,shows regularity variation during the processing of De-enzyming of the fresh tea leaves.(4) Take Zhenmei Tea and Zhuyeqing Tea as example,the method that can quantize the quality of appearance of tea using computer vision image processing technology was explored.The results showed that the color parameters of R was the best one for evaluating the degree of Zhenmei Tea in RGB color model;The pixel value of tea image can reflect the area of tea,and the pixel value of tea image will become less when the degree of Zhenmei Tea became low,the correlation coefficient is 0.904**;the standard deviation of green channel and brigtness channel can reflect the uniform degrees of Zhenmei Tea well; the coefficient of variation of the parameters of R,L,a and the pixel value were better ones for judging the quality of Zhuyeqing Tea.(5) A new method of mensuration disease area and the percentage of diseases area of tea leaves with computer,which should using Photoshop software and scanning many leaves,has been established,and the further investigation also has been fulfilled.The results showed that when scanned with resolution 150dpi,there were be short time to scanning and less disk space been occupied,and the effect as same as scanned with high-resolution.The results of mensuration disease area and percentage of diseases area of tea leaves using the new method also as same as the results using artificial mensuration, and the efficiency of the new method is much higher than the efficiency of artificial method. |
PDF Full Text Request