基于全局和局部特征融合的图像匹配算法研究

针对移动机器人视觉同时定位与地图构建过程中图像处理速度慢以及特征点匹配实时性和准确性差的问题,提出基于颜色特征和改进SURF算法融合的图像匹配算法。首先,采用颜色特征对图像序列进行粗匹配,选取与测试图像最相近的5幅图像作为待匹配图像;其次,改进SURF算法,用Krawtchouk矩对采用Hessian矩阵获取的关键点进行描述,计算关键点的梯度方向和幅值,得到新的特征向量,对待匹配图像提取改进SURF特征再与测试图像进行精确匹配,得到最佳匹配图像,此匹配算法提高了移动机器人图像处理的速度和精度。实验结果表明,改进算法的误匹配率降低10%左右,程序运行时间减少,在可靠性得到保证的同时适应于实时性应用。

Investigation on the image matching algorithm basedon global and local feature fusion

Aiming at the problems of slow image processing speed, poor real-time capability and accuracy of feature point matching in mobile robot vision-based SLAM. The paper proposes a novel image matching method based on color feature and improved SURF algorithm. Firstly, color characteristics is adopted to roughly match the image sequences, and five images most similar to the test image are selected as the image to be matched. Then, the SURF algorithm is improved, the Krawtchouk moment is adopted to describe the feature points obtained using the Hessian matrix, and the gradient direction and amplitude of the feature points are calculated, and then, the new feature vector is obtained. In the image matching process, the improved SURF features are extracted from the image to be matched, which are matched with that of the test image precisely. Finally, the best matched image is obtained. This matching method improves the speed and accuracy of the mobile robot image processing. Experiment results show that the error matching rate of the proposed algorithm is decreased by about 10%, and the program running time is reduced. What''s more, this method is suitable for real-time applications while guaranteeing the reliability.