【作者】 叶祖霈； 唐林波； 赵保军；

【Author】 YE Zupei;TANG Linbo;ZHAO Baojun;School of Information and Electronics,Beijing Institute of Technology;

【机构】 北京理工大学信息与电子学院；

【摘要】 提出了一种基于二阶灰度川页序的局部特征描述子改进算法。该算法只需要使用与现有算法相同的采样点,通过计算二阶灰度顺序来挖掘更深层次的关联信息,并在假设的图像变换中能够保证特征的稳定性。实验结果表明,提出的算法在图像匹配中的性能超过了现有的先进算法,并且在各种图像变化发生时仍能够保持较高的准确率,验证了二阶灰度顺序对增强局部特征描述子分辨能力的有效性。更多还原

【Abstract】 This paper proposed an improvement algorithm of local image descriptor based on second intensity order.The algorithm uses identical sample points as existing algorithms and mines deep-seated information by computing the second intensity order.Also it can maintain robustness under given image transformations.Experimental results on image matching demonstrates that the proposed algorithm outperforms the existing state-of-the-art algorithms and can keep high accuracy under varieties of image transformations,which shows the effectiveness of the second intensity order on improving the discriminability of local feature descriptor.更多还原