本文提出一种基于核方法的下视等分辨率景象匹配算法. 通过模拟电荷吸引模型, 提出了计算不等维高维数据相似度的SNN 核函数. 将图像中的特征点映射到径向基向量(Radial basis vector, RBV) 空间, 利用SNN 核函数计算两个特征点集的相似度及过渡矩阵. 利用置换测试模块来增强SNN 核的稳定性, 以确保输出解的可靠性. 实验证明, 基于SNN 核的景象匹配算法对图象畸变、噪声干扰与信号缺失具有很强的鲁棒性, 并可保证高精度与高实时性.

-This paper presents a method based on kernel resolution of the next scene, as the other matching algorithms. Attracted by charge simulation model for computing the data ranging from high-dimensional similarity dimension SNN kernel function. The image feature points mapped to the radial basis vector (Radial basis vector, RBV) space, calculated using SNN kernel similarity of two feature point sets and the transition matrix. permutation test module to enhance the stability of the nuclear SNN to ensure the reliability of the output solution. Experiments show that SNN-based scene matching algorithm kernel image distortion, noise and signal loss are highly robust, and can ensure high precision and high real time.