网络回溯的算法模型改进及电网应用

Based on the Network Back Recognition Technology and Its Application Research

对称正定(Symmetric Positive Definite,SPD)矩阵可以通过计算图像集合的协方差矩阵建立图像集合的模型。文中基于网络回溯和对称正定进行了图像加密识别的研究。通过对标准数据库集分析实验得出:提出的方法获得了较高的识别精度,将流形上的点投影到高维的核空间,应用核鉴别分析进行分类。本文使用无监督的方法获得的投影矩阵达到降维的目的 ,性能上较同类的方法有较大的优势,但是识别精度仍有待提高。研究结果对于识别进行的优化具有一定的参考价值。

Symmetric Positive Definite （Symmetric Positive Definite, SPD） matrix can be very good to encode image information. In this paper, based on the network back and symmetric positive definite to encrypt the image recognition research. Through analyzing standard database set experiment, puts forward the method to obtain the higher recognition accuracy, the points on the manifold projection to high dimensional kernel space, using nuclear identify classification analysis. This article Uses unsupervised method to obtain the projection matrix to achieve the purpose of dimension reduction, performance on a similar method has great advantage, but still needs to improve identification precision. The results of the study identify the optimization has a certain reference value.