中文摘要：

针对移动机器人定位过程中视觉图像特征点提取与匹配数量较多,边缘特征点不稳定的问题,提出基于改进离散Gaussian-Hermite矩的SURF图像匹配算法。利用双树复小波变换将图像分解为低频与高频部分,将低频部分作为改进算法的输入图像。通过采用3D非极大值抑制求取图像不同尺度下的特征点,计算图像Gaussian-Hermite矩,重新确定Hessian矩阵,检测稳定边缘特征点,定义新的特征描述向量。将改进算法与自适应粒子滤波定位算法相结合,实现移动机器人在室内环境中的视觉定位。实验结果表明：改进算法配准精度高于SURF算法,不稳定特征点提取数量相比于原算法约减少9%,匹配率得到进一步提升。

英文摘要：

Aiming at the problems of large amounts of characteristic point extraction and matching for visual image and the unstability of the edge characteristic points,a SURF image matching algorithm based on modified discrete Gaussian-Hermite moment is proposed. The dual-tree complex wavelet transform is adopted to decompose the image into low frequency part and high frequency part,and the low frequency part is used as the input image of the improved SURF algorithm. The 3D non-maxima suppression algorithm is adopted to determine the feature points under different scales,calculate the image Gaussian-Hermite moment,re-determine the Hessian matrix,detect the stable edge characteristic points and define the new characteristic description vectors of the interest points. The improved algorithm is combined with the adaptive particle filter localization algorithm,and the visual localization of the mobile robot in room environment is realized. The experiment results show that the registration accuracy of the improved algorithm is higher than that of the SURF algorithm;compared with the original algorithm,the number of the extracted unstable feature points is decreased by about 9%,and the matching rate is further enhanced.