Nutrition Diagnosis Of Phyllostachys Pubescens Forest Based On Leaf Color Characteristics

The traditional method to determine the content of chlorophyll to diagnose the nutritional stat us of plants,the workload is heavy and complicated,the use of modern scientific and technolo gical means,whether spectrometer or chlorophyll meter,also requires expensive equipment,prod uction practice is difficult to be promoted,this paper studies the method of plant nutrition diag nosis based on Digital image processing technology.In this experiment,mature leaves of annual Phyllostachys pubescens in field environment were selected as the research object.The experi ment was set up in different nutrient levels of Phyllostachys pubescens Forest in Suichang coun ty.The widely used mobile phone equipment（Huawei P30 as an example）was used to obtain images of Phyllostachys pubescens leaves The main results are as follows:（1）Extraction of color features of Moso BambooThe direct shooting of blades would be affected by the definition of the picture,the sun’s height angle and the shooting parameters.On the basis of a great deal of research,（Take Hua wei P30 as an example）,it is clear that the shooting height is 80CM from the blade.The shoo ting parameters are:ISO value 100～125,S（shutter speed）is fixed at 1/60,F（focal length）i s 27mm,WB（white balance）,automatic white balance)（AWB）,shooting time is 9.30 am-15.00pm（clear and calm）,shooting background is black,etc.From the quality of the acquired ima ge,the image is clear and has low noise and less interference in the target area,which reduces the difficulty of later image processing,reduces the time of machine operation,and also provi des an important technical reference for related research work.Because of the influence of une ven illumination and shadows on the leaf images of Phyllostachys pubescens obtained under nat ural illumination conditions,this study adopts the method of normalizing the extracted R,G an d B mean values to eliminate the influence of illumination and shadows in mathematical sense,and carries out mathematical transformation to improve the correlation between color features a nd chlorophyll content.（2）Construction of chlorophyll content model of Phyllostachys pubescens leavesBased on the correlation model between leaf color characteristic parameters and chlorophyll content,it has a great correlation with color characteristic parameters 2r-g-b,r,r/（g+b）and Ch la+Chlb（chlorophyll a+chlorophyll b）.The color characteristic parameters 2r-g-b,r/（g+b）,g/r,g-r,R have great correlation with Chla（chlorophyll a）,and the color characteristic parameter s 2r-g-b,r/（g+b）,r,r-b have a great correlation with Chlb（chlorophyll b）.Based on the optimi zed color characteristics and chlorophyll content,one-dimensional linear regression model and m ultiple linear regression model are constructed.After model verification,the average relative err or of the selected three multiple linear regression models is between 7%and 13.4%,and the d etermination coefficient R² is 0.4341～0.7905,which can better estimate chlorophyll content.（3）Construction of discriminant model for different nutrient levels of Phyllostachys pubescens ForestThe color characteristic parameters g,b,2r-g-b,b/g were selected by stepwise discriminant analysis,and then three discriminant classification functions were established by Fisher canonic al discriminant analysis,Y₁=362305.125x₁-51373.028x₂+57480.102x₃+154281.990x₄-111584.072,Y₂=365571.599x₁-54422.623x₂+57838.793x₃+156729.692x₄-113476.440,and Y₃=365464.460x₁-54728.946x₂+57689.673x₃+156812.614x₄-113394.292.The model was verified by knife cutting method and back generation estimation method,and the discrimination accuracy was 100%,which could bett er treat wool with different nutrition levels.