Research On Image Recognition Based On Convolution Neural Network

With the development of the Internet,image classification technology become more and more mature and is widely used,can be large-scale graphics classification according to a certain algorithm to improve the efficiency of information classification and utilization,but also can be a variety of image features extracted,there is a great help for the later image processing.In this paper,we first introduce several commonly used image classification methods and their research status.At the same time because the traditional methods of small sample image can only deal with small sample data set classification,image more number of data sets to handle,so that the image classification problem to deal with a large amount of data using convolutional neural network,and to do the following work:(1)The basic theory of convolutional neural network is presented in this paper,the artificial neural network and BP neural network are introduced,starting from the structure and characteristics of convolutional neural network,in-depth study of the principle of operation of each different information processing layer,summarizes the characteristics of convolutional neural network,the weights of the convolutional neural network update was deduced,prepare for later experiment.(2)Classification and prediction in image data set CIFAR-10 as the research object,experimental verification of the classification of the ability to handle large data sets of the convolutional neural network,and gives the classification accuracy;the effects of various factors on the prediction effect of dataset classification are studied.(3)The convolutional neural network and support vector machine(SVM)is connected to complete image classification.Convolution neural network is a kind of deep learning algorithm,for image feature extraction,and then in the last layer support vector machine(SVM)is used for classification.By experimental verification,image classification effect is achieved.