为了实时准确地监控网络上的不良图像,本文提出了一种新的基于SVM的不良图像检测方法,首先融合不良图像的物理特征构造SVM分类器,将在应用层网关截获的图像信息提交给分类器进行识别,对识别结果为不良的图像进行IP地址屏蔽。实验结果表明本文提出的基于SVM的不良图像检测方法可成功用于网络上不良图像的监控。 In order to monitor the bad images on the network in real time, this paper proposes a new method of image defect detection based on SVM. Firstly, the SVM classifier is constructed by combining the physical features of bad images, and the image information captured by the application gateway is submitted to the classification Identification of the device, the recognition result is poor image IP address mask. The experimental results show that the proposed SVM-based image defect detection method can be successfully used to monitor the bad images on the network.