卷积神经网络在图像分类上的应用综述

Abstract

近些年来，图像分类在模式识别领域占据着重要的地位。迅速增长的图像数据对于图像信息的分析与处理提出了新的要求。卷积神经网络应运而生，以其强大的图像识别分类能力被广泛的应用于各种图像分类系统，并取得了十分显著的效果。本文首先，回顾了图像分类技术的发展历程，介绍了图样空间与特征空间的分类方法以及图像特征提取方法。其次，介绍了卷积神经网络在图像分类问题上的研究情况。最后，总结卷积神经网络在图像分类问题上存在的问题，以及未来的发展方向。 In recent years, image classification occupies an important position in pattern recognition. The rapid growth of image data for image information analysis and processing put forward new re-quirements. Convolution neural network came into being, with its powerful image recognition classification ability is widely used in a variety of image classification system, and achieved very significant results. Firstly, this paper reviews the development of image classification technology, and introduces the classification methods of feature space and feature space and the image fea-ture extraction methods. Secondly, the research of convolution neural network on image classi-fication is introduced. Finally, the problems existing in the image classification of convolution neural networks are summarized, and the future development direction is summarized.