基于卷积神经网络的视频图像失真检测及分类  

Video image distortion detection and classification based on CNN

作　　者：邬美银 陈黎[1] 田菁[1] Wu Meiyin;Chen Li;Tian Jing

机构地区：[1]武汉科技大学计算机科学与技术学院智能信息处理与实时工业系统湖北省重点实验室,武汉430065

出　　处：《计算机应用研究》2016年第9期2827-2830,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61375017);湖北省高等学校优秀中青年科技创新团队计划资助项目(T201202)

摘　　要：为了检测不同失真类型的视频图像,实现对失真视频图像的分类处理,提出一种基于卷积神经网络的视频图像失真检测及分类方法。将视频图像分割成较小的图像块作为输入;然后利用卷积神经网络主动学习特征,引入正负例均衡化和自适应学习速率减缓过拟合和局部最小值问题,由softmax分类器预测图像块的失真类型;最后采用多数表决规则,得到视频图像的预测类别。采用仿真标准图像库(LIVE)和实际监控视频库对该方法进行性能测试,前者的总体分类准确率达到92.22%,后者的总体分类准确率达到92.86%。整体的分类准确率高于已有的其他三种算法。引入正负例均衡化和自适应学习速率后,CNN的分类准确率得到明显提升。实验结果表明,该方法能主动学习图像质量特征,提高失真视频图像分类检测的准确率,通用于任意失真类型的视频图像分类检测,具有较强的鲁棒性和实用性。To detect different kind of distortion in video images, and process the distortion video images according to the category, this paper proposed a detection and classification method of distortion video images based on convolutional neural network (CNN). It took the small patches which were segmented from the video images as input. Then it used CNN to learn image features, and predicted the class of the patches with a softmax classifier. To reduce overfitting and local minimum, it introduced the positive and negative sample equalization and an adaptive learning rate. Finally, it adopted the majority votingrule to decide the class of video images. It conducted the performance evaluation for the proposed algorithm on the standard image database( LIVE) and the real-world surveillance video database. The total classification accuracy was up to 92.22% on the former, and 92. 86 % on the latter. The overall classification accuracy was higher than the other three algorithms. With the introduce of positive and negative sample equalization and an adaptive learning rate, CNN had better performance. Experimental results show that the proposed method can learn image quality features by itself, and improves the classification accuracy. It has good robustness and practicability, and is suitable for any kind of distortion video images.

关 键 词：卷积神经网络 特征学习 视频图像失真 分类检测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]