Research On Image Recognition Model Of Vegetation Based On Location Based Service

Vegetation image recognition is one of the important research contents of intelligent recognition.It plays an important role in agricultural planting,vegetation science research,rare vegetation protection,ecological civilization construction,and vegetation culture communication.Nowadays,many location-based Service applications also incorporate image recognition technology to enhance the user experience.However,with the continuous development of shooting technology,more image details will be captured and displayed in front of our eyes.The angle of the picture,the resolution,the area where the vegetation grows,the various forms of vegetation and the shape of different vegetation in different growth stages.it brings a huge challenge to the traditional image classification algorithm.The image recognition function is an important module in the campus vegetation application based on LBS.It is obviously not enough to rely on the optimization model itself to improve the recognition accuracy.Therefore,based on the above problems,the topic of this paper is how to construct a scientific vegetation image recognition model based on the idea of geographic information system.In this paper,four image classification algorithms,including support vector machine,K nearest neighbor classification algorithm,convolutional neural network,and transfer learning,are studied before the image classification model is constructed.After analyzing the advantages and disadvantages of each algorithm,the two common algorithms of support vector machine and convolutional neural network are selected to compare the experimental results of the image classification model.Finally,via the experimental results,the classification model based on convolutional neural network has high accuracy in image classification and is suitable for the campus vegetation identification task studied in this paper.Then,by optimizing the convolutional neural network model and combining the ideas of the geographic information system,the classification model is integrated into the position and time to identify the initial classification results,which can further improved the recognition accuracy.The work of the specific paper is as follows:(1)Build a picture crawler program,crawl the Internet's rich vegetation image resources by parsing the principle of obtaining the image address on the webpage,which is also the data source method for constructing the vegetation image dataset.(2)The image classification model based on CNN and SVM is designed respectively.The vegetation image recognition model is constructed by analyzing the experimental results and the advantages and disadvantages of the algorithm and selecting the convolutional neural network which is suitable for the vegetation image classification task studying in this paper.(3)Based on the idea of geographic information system,the location and time factors are integrated into the LBS-based campus vegetation recognition program to correct the results of recognition errors,it solved the problem of "guess" caused by probability of image classification model and greatly improved the accuracy of recognition.(4)The image classification model based on CNN is optimized and embedded into the location-based service application.The identification strategy of the initial classification result based on location and time factors is designed,which makes the application have certain practical value.In this paper,the convolutional neural network is selected to train the vegetation image recognition model,and the model optimization is used to further improve the recognition accuracy.Then the model is embedded into the location-based service-based application.The misjudgment results were corrected by time and location identification strategies to make the model more scientific.