Research On Classification And Recognition Of Rose Image Based On Neural Network

As an important industry in our country,roses have extremely important economic and ornamental value,and they have become more and more popular among people.However,due to the influence of factors such as man-made and planting environment,even the same variety of roses will appear different levels of classification.At present,the classification of roses is mainly carried out manually,which is easy to cause inefficient problems such as misclassification,omission,and duplication of work.In order to realize the recognition and classification of roses from manual classification to machine autonomous,this paper classifies and recognizes roses based on the image feature data of roses and neural network algorithms.In this paper,196 images of yellow roses with 5 grades of the same species were obtained by manual shooting.After the collected rose images are subjected to operations such as data enhancement and image naming,an image database is constructed.Under the python language and tensorflow deep learning tool,build an artificial neural network model and a convolutional neural network model,and then send the image data set into the artificial neural network and the convolutional neural network respectively.The experimental results show that the recognition rate on the test set is relatively low,and the difference between the recognition rate on the training set is relatively large.In order to solve this problem,this paper extracts 12 feature variables from three aspects of color feature,geometric feature,and texture feature,and uses the component coefficient and scoring coefficient in the principal component analysis method to evaluate and analyze the parameters of the 12 feature variables.Then this paper will use the extracted feature data again based on the artificial neural network model to classify and recognize the yellow roses of the same species.The experimental results show that the recognition rate on the test set is significantly improved,and the difference between the recognition rate on the training set is very small.Finally,100 trainings were performed under the three models,and the average recognition rate under the three models was obtained.The results show that compared with the other two models,the artificial neural network model based on feature data has a significantly faster running speed and a significantly higher average recognition rate.