Study On The Identification Of Chrysanthemun In Natural Environment

As the hometown of tea,China is a big country of selling tea drinks in the world.Nowadays,Hangzhou White Chrysanthemum is more and more popular because of its health benefits.The birthplace of Hangzhou White Chrysanthemums is Tongxiang city of Zhejiang Province,where output accounts for 70% of nation-wide total output.According to the different flowering periods,Hangzhou white Chrysanthemum is mainly divided into fetal-Chrysanthemum and mature-Chrysanthemum.The former is the Hangzhou white Chrysanthemum which is in the beginning of bloom,and the latter is Hangzhou white chrysanthemum which is in the end of bloom.Compared with mature-Chrysanthemum,fetal-Chrysanthemum has higher medical value and market price.Thus,the primary target of picking is the fetal-chrysanthemum during the flowering period.At present,the whole process of harvesting is carried out by manual operation,which is a time-consuming and labour-intensive procedure.Due to the short flowering time,many fetal-Chrysanthemums would not be picked in time.It is necessary to adopt automatic picking method to improve the picking efficiency and reduce the labor force.And the identification of fetal-chrysanthemum is one of the key techniques of automatic picking.With fetal-chrysanthemum which lived in natural environment as the subject.The key of segmentation and recognition of fetal-chrysanthemum is studied on the basis of domestic and foreign related research.The machine vision technology is used to segment the fetal-chrysanthemum from the complex background.It will prepare for the final realization of automatic picking.The main contents of the research are as follows:1)In order to improve the quality of collected images,the factor which influenced the recognition of fetal-Chrysanthemum were analyzed.According to the theory of machine vision,the hardware and software of recognition system for fetal-Chrysanthemum were established.The intensity of light at different time periods was measured under the condition of front lighting and back lighting,and the growth posture of fetal-Chrysanthemum was observed to determine the appropriate image acquisition scheme.2)Several classical image segmentation algorithms were used to remove background regions.There are various background elements in the natural environment,such as branches,leaves,soil and withered grass.In order to extract complete target regions,a color-texture segmentation method was proposed.It mainly consists two parts.Fist,the 2G-R-B features were modified into 4G-R-B to remove dead grass and branches.Second,due to the texture difference between soil and flowers,the flowers target can be completely separated from the background by texture feature.3)In order to solve the problem of separating overlapping Hangzhou White Chrysanthemum,three algorithms were studied.They are watershed algorithm,edge detection algorithm and K-means clustering algorithm.The experimental results shows that the average error rates of three algorithms are 5.116%,6.336% and 3.813% respectively.The average times of three algorithms are 1.9490 s,1.2595 s and 2.310 s.It was observed that the method based on K-means clustering algorithm can better adapt to different kinds of overlapping Hangzhou White Chrysanthemum.4)A method based on shape feature was proposed to distinguish the fetal-chrysanthemum from Hangzhou White Chrysanthemums which is in others flowering periods.The area,length ratio and roundness of Hangzhou White Chrysanthemums were extracted.The experimental results show that the average accuracy of the method is 95.651%.