基于GAN和CNN-ELM的监控图像智能修复及检测方法

针对目前石化危险品装车过程中海量监控视频图像人为处理效率低下、模糊图像识别率低等问题,提出一种基于生成式对抗网络(GAN)和卷积神经网络(CNN)与极限学习机(ELM)相结合的监控模糊图像智能修复及检测方法.首先,使用深度学习网络作为 目标检测框架,利用GAN网络中生成器与判别器间的零和博弈对模糊图像进行复原,得到清晰完整的作业图像;其次,利用CNN自适应学习图像特征的能力,对修复后的图像进行自主特征提取;最后,将提取的图像特征输入ELM分类器中进行目标识别与分类,判断作业过程是否存在违规行为.试验结果表明:所提方法图像修复速度快,视觉效果自然,且目标识别准确率高,具有很好的泛化能力.

Intelligent Restoration and Detection Method of Monitoring Image Based on GAN and CNN-ELM

Aiming at the problems of low artificial processing efficiency and low recognition rate of fuzzy image in the process of loading petrochemical dangerous goods, an intelligent repair and detection method of monitoring fuzzy image based on the combination of generative adversarial network (GAN), convolutional neural network (CNN) and extreme learning machine (ELM) is proposed. Firstly, using the deep learning network as the target detection framework, the fuzzy image is restored by using the zero sum game between the generator and the discriminator in the generative adversarial network to obtain a clear and complete job image. Secondly, using the ability of convolutional neural network to adaptively learn image features, the autonomous features of the repaired image are extracted. Finally, the extracted features are input into the extreme learning machine classifier for target recognition and classification to judge whether there are violations in the operation process. The experimental results show that the proposed method has fast image restoration speed, natural visual effect, high accuracy of target recognition and good generalization ability.