基于SCCH特征描述子的图像匹配算法

摘要

针对局部特征匹配算法面临的鲁棒性和实时性难以兼顾的问题,该文提出了一种基于带符号对比上下文直方图(SCCH)特征描述子的图像匹配算法.用Harris算子在高斯金字塔图像上提取多尺度特征点以减少所需处理的数据量,利用特征点邻域的区域灰度差异均值构建特征描述子,降低特征描述子的生成复杂度和维度,保留灰度差异均值的正负性信息以增强特征描述子的鲁棒性和可区别性,用特征描述子间的绝对值距离作为相似性度量以减少特征点匹配的计算量.实验结果表明,该文算法不仅对图像尺度缩放、旋转、模糊、亮度变化、较小视角变化保持不变性,而且匹配速度较快.%In order to solve the problem that it is difficult to balance the real-time performance and robustness in image matching using local feature, an image matching algorithm based on Signed Contrast Context Histogram (SCCH) feature descriptor is presented. Multi-scale feature points are extracted with Harris operator in Gaussian pyramid images to reduce the data for processing. Feature descriptor is built with the means of the differences of gray value in the sub-regions of feature point neighborhood, which not only decreases the complexity of building descriptor and the dimensions of descriptor, but also enhances the robustness and distinctiveness of descriptor. Furthermore, the absolute distance between descriptors is used to match feature points as a similarity measurement to lessen the computation. Simulation results indicate that the proposed algorithm keeps invariant in the case of image zoom, rotation, blurring, luminance varying as well as smaller view angle changes, and its matching speed is faster.