卷积神经网络在图像处理中的应用

吉林建筑大学 电气与计算机学院,长春 130118

卷积神经网络; 图像处理; 图像识别

Application of convolutional neural network in image processing
YAN He,GUO Xiu-juan

School of electrical and computer science,Jilin Jianzhu university,Changchun 130118,China

convolutional neural network; image processing; image recognition

备注

卷积神经网络相比普通的神经网络能够大量减少训练集,并且识别准确性也能大幅提高,目前在图像处理和图像识别等领域有非常广的应用.本文详细阐述卷积神经网络,包括卷积层、激活层、池化层、全链接层,并利用TensorFlow做一个对比,展示卷积神经网络在图像处理中的过程.

The Convolutional neural network compared to common neural network can reduce the training set,and accurate identification can also greatly improve,in areas such as image processing and image recognition has a very wide range of applications,this paper first introduces convolution neural networks,including the convolution,the activation layer,pooling layer,link layer,finally make a simple comparison between using TensorFlow,show the convolutional neural network in the process of image processing.