一种基于双网络模型的图像识别新方法

针对基于复杂网络图像识别方法建模复杂度过高问题，提出一种基于双网络模型的灰度图像识别新方法。首先将像素点作为复杂网络节点，基于灰度乘积构建图像的结构平衡网络模型以及基于欧氏距离构建复杂网络模型，然后分别计算两种网络模型的拓扑特征参量，形成最终的图像特征识别参量。相比现有基于复杂网络的图像识别方法，该方法在理论上能够降低图像建模复杂度，提高图像识别速度。使用 YALE 人脸数据库进行仿真对比实验，结果表明，该方法的图像识别速度为传统复杂网络方法的 35%，正确率提高了 4%。

Novel Dual Complex Network Based Image Recognition Method

Aiming at the influence of the number of nodes on complex network face recognition methods，we propose a novel image recognition methodology based on dual complex network. First，by viewing the pixels of the grayscale image as the nodes of the complex network，a structure balanced network model is constructed by pixel-gray product and complex network models are constructed respectively based on the Euclidean distance among the nodes. Then，the topological measures of the two networks are calculated and the corresponding image recognition features are extracted from the derived topological measures. Compared with the complex network based image recognition approach，the proposed method is able to promote the image recognition speed，for it is much easier than the existing in constructing the network model of the image. In order to evaluate the identified advantages of the proposed approach，contrast tests are conducted based on the classical YALE face database. The experimental results show that the method speed of the proposed approach is 35% of traditional complex network，and the accuracy rate is improved by 4%.