针对人脸图像复原问题,提出了基于偏最小二乘法回归(Partial Least Squares,PLS)的超分辨率复原技术与小波双立方配比插值放大结合的算法。对低分辨率图像进行小波分解后得到的3个高频子图像插值,然后用小波逆变换进行重建。同时对低分辨图像进行双立方插值,插值后的图像与小波重建图像进行融合,提取融合图像特征作为训练样本,应用偏最小二乘法回归进行样本训练,最后得到复原高分辨率图像。试验结果表明,基于偏最小二乘法小波双立方配比插值的图像复原算法比传统插值算法(最近邻插值、双线性插值、三次样条插值)在人脸图像复原方面效果好。
The next generation protocol IPv6 brings the new challenges to the information security. This paper presents the design and implementation of a network based intrusion detection system that support both IPv6 protocol and IPv4 protocol. This system＇s architecture is focused on performance, simplicity, and scalability. There are four primary subsystems that make it up： the packet capture, the packet decoder, the detection engine, and the logging and alerting subsystem. This paper further describes a new approach to packet capture whose goal is to improve the performance of the capture process at high speeds. The evaluation shows that the system has a good performance to detect IPv6 attacks and IPv4 attacks, and achieves 61% correct detection rate with 20% false detection rate at the speed of 100 Mb·s^-1.
