一种鲁棒的基于图像对比度的局部特征描述方法

摘要

A robust local feature description method based on image contrast is proposed, which is called Independent Elementary Contrast Histogram (IECH) descriptor. First, the contrast value between each pixel in the local region and the pixel which is chose by random sampling is computed. Second, the local region is divided into 32 sub-regions starting from the dominate orientation in the log-polar coordinate system, and a 2-bins contrast histogram is calculated in every sub-region. Finally, the histogram vector is normalized to create the 64-dimensional IECH descriptor. By comprehensive comparison with other descriptors, the results indicate that the proposed descriptor is competitive with the performance of SIFT descriptor, while getting higher descriptor building speed and lower descriptor dimension. Moreover, the proposed method possesses a superior performance compared to the Contrast Context Histogram (CCH) descriptor with the same time complexity and descriptor dimension, and it is more suitable for real-time applications.%该文提出一种鲁棒的基于对比度的局部特征描述方法，即独立元素对比度直方图(Independent Elementary Contrast Histogram, IECH)描述子。首先计算特征区域内各像素与被随机采样像素间的对比度值。然后，在极坐标下以特征主方向为基准，将局部特征区域分割成32个子区域，分别统计2维正负对比度直方图。最后，对统计结果进行归一化处理，产生64维的IECH特征描述向量。实验结果表明，该方法在保持与SIFT相当的匹配性能的同时，具有更快的特征生成速度与更低的特征维数。相比于具有相同时间复杂度与特征维数的对比度上下文直方图(CCH)方法，该方法描述子的性能有了明显的提高，更适合在实时应用中使用。