Design And Development Of Plant Specimen Collection Information System

The plant specimen collection information system is a joint development project of the Information Engineering of Northwest A&F University and the Xi'an Middle School Biology Teaching and Research Group.The existing plant specimen information system pays more attention to data analysis and management of specimen information,and many UN family-level specimen databases are difficult for ordinary people to use.The general plant image recognition system has a single function.At present,there is no professional and comprehensive system for providing specimen entry management and plant identification.In view of the deficiencies of the above systems,this study is based on the software requirement description of Xi'an Middle School Biology Teaching and Research Group.It aims at the collection of Qinling specimens for middle school students and develops a mobile phone APP to achieve registration,personal information management,plant specimen upload,notification information management,and specimens.Search and plant image recognition five modules function.In this paper,the SAAK model and the large margin classifier based on affine hulls algorithm(LMC)are used to extract plant image feature information and image classification.The content of this article expands from the following aspects:(1)Based on the software requirements of Xi'an Middle School biology teaching and research group,a collection and information system of plant specimens was designed and implemented.The system adopts the C/S architecture,uses HTML5,CSS and JavaScript to develop the client,simplifies the UI design,sets up the Python language,Django framework and MySQL data backend,and implements data interaction processing.The Django framework is an MTV framework model consisting of a model,a view,and a template,which reduces the degree of coupling and realizes the separation of services in each part.(2)The collection activities of Qinling specimens are prone to a situation in which the network environment is not smooth,resulting in the system being unable to use normally.In order to make the user login and specimen upload functions independent of the network conditions,the system designs two methods for uploading and storing data.When the network is unblocked,users submit data to interact with the database through the data interaction end.When it is impossible to connect to the network,serialize the sample information in the local file of the system,and automatically upload it when the network is good,and realize the function of uploading without the network sample.Deserialize all users stored locally to obtain a JSON object,query whether the user information exists,and realize that the user does not log on the network.(3)In order to realize the function of plant image recognition,two feature extraction and classification algorithms are studied and trained to classify the PlantCLEF dataset.The first is the SAAK model combined support vector machine(SVM)algorithm.Using the three kernel functions in SVM: the recognition rates of Gaussian kernel function,linear kernel function,and polynomial kernel function are 69.82%,60.96%,and 64.52%,respectively.The second is the largest edge classification algorithm based on affine packets proposed by Hasan Serhan Yavuz et al.The accuracy of LMC classification algorithm is between RBF SVM and Polynomial SVM,which is 66.34%.Therefore,the SAAK model is integrated with the Gaussian kernel function SVM integrated algorithm to realize the plant image recognition function.