基于方向角的特征配准算法

Image feature points match algorithm based on orientation angles

图像特征点匹配是计算机视觉领域的一个瓶颈问题,为了提高匹配精度,提出了一种基于方向角的特征配准算法。该算法采用Harris算子检测角点,根据图像梯度变化计算特征点的方向角,然后将待匹配特征点对间的方向角距离转化为权值对代价函数进行加权,并对特征点进行分析,改进了代价函数。实验证明,本文算法的匹配精度高于传统的归一化互相关(NCC)匹配,且具有一定的旋转不变性和抗噪性。

Image feature points match is a crucial problem in computer vision.In order to improve the match precision,feature points match based on orientation angles is proposed.This method adopts Harris arithmetic operators to detect corners.Orientation angles of feature points are calculated according to image grads.Then the distances between orientation angles of feature points are translated into weighted coefficients,and the cost function is weighted by them.We analyze the feature and improve the cost function.E...