摘要
为了提高机器人在环境识别中的实时性和准确性,在分析了SIFT描述算子性能及原理的基础上,对SIFT算法在实时性方面改进。在特征点进行欧氏距离的匹配的时候加入一个阈值,将那些不匹配的点可以直接滤除;在建立图像金字塔的时候没必要一直进行到最后一组足够小为止,可以通过实验得到第4组的时候特征点的总数就不再增加了,不需要一直计算下去从而节省时间。仿真结果也证明了其有效性。
In order to raise the real-time and accuracy of environment recognition for robot,the SIFT algorithm in real-time is improved by analyzing the performance and principle of SIFT descriptors.A new threshold is added to the process of the match between the two features,in order to exclude directly the features which are not matched.It is not need to continue until the last group enough small when the image pyramid is established,and though the experiment it is found that the total number of features will no longer be increased over the first 4 group.The simulation show its effectiveness.
出处
《电气传动自动化》
2012年第2期44-47,共4页
Electric Drive Automation
