Abstract: SIFT（Scale Invariant Feature Transform） algorithm has the characteristics of scale，rotation and illumination invariability，so it is widely applied to the field of mobile robot vision positioning.Because of the change of special maker image size or background image size has a great influence on position calculation time，in order to improve the mobile robot’s real-time vision positioning capability，the size of maker image and the back background image are changed respectively，and the corresponding real-time test of image matching is put forward.The test results show that when the image size reduces 40%～60%，the real-time character rises 30%～40%，and the characteristic points reduce 20%～30%，on the premise of matching precision，the position real-time capability increase markedly.Therefore a suitable size of image can improve the position real-time capability with a high matching precision and this method also gives guidance for the other similar use.

Key words: Scale Invariant Feature Transform（SFIT） algorithm, mobile robot, image matching, vision orientation

摘要： 尺度不变特征变换（Scale Invariant Feature Transform，SIFT）算法由于具有尺度、旋转和光照不变等特性，被广泛应用到移动机器人视觉定位领域。由于定位计算中使用的特定标示物图像及背景图像的尺寸对定位计算时间影响较大，为提高移动机器人视觉定位的实时性能，分别改变待识别图像以及背景图像的尺寸，进行了图像匹配实时性测试实验。实验结果表明，特定标示物图像和背景图像的尺寸缩小到40%～60%时，移动机器人视觉定位实时性提高30%～40%左右，特征点数减少20%～30%左右。在满足定位精度要求的情况下，大大提高了定位的实时性。由此可知，通过合理选择图像的尺寸不仅可以满足匹配精度，还能提高移动机器人的视觉定位效率，对于类似该实验使用的移动机器人视觉定位具有广泛的指导意义。

关键词: 尺度不变特征变换算法, 移动机器人, 图像匹配, 视觉定位