基于改进的Harris角点检测在岩心图像拼接中的应用

The Application of Improved Harris Corner Detection in Core Image Stitching

提出了采用基于改进的Harris算子的自动拼接技术,以获取高质量岩心扫描全景图。自动拼接的关键是特征点的提取和匹配。Harris角点检测算法稳定、计算简单、不受光照、旋转等影响,但检测尺度单一;SIFT特征描述算子以其稳定性而著称,但要得到128维的特征描述向量,计算复杂而不易实现。本文结合二者的优点,采用高斯多尺度改进Harris算子检测岩心图像点特征,并采用简化SIFT描述点特征,然后用欧式距离进行特征匹配。实验结果表明,对大多数岩心图像,采用该算法能获得拼接较准确的岩心全景图,在以往基础上提高了岩心图像拼接系统的准确性,具有一定实用价值。

This paper proposes Automatic Image Stitching technology on basis of improved Harris operator to obtain high quality core scanning panoramic images.The key in automatic stitching is feature point extraction and matching.Harris Corner Detection Algorithm is stable,easy to compute,and unaffected by light and rotation,but a single scale;SIFT feature description operator is known for its stability,but it is not easy to obtain the 128-dimensional characterization vector,for its computational complexity.Combining the advantages of both of them,this paper uses Gaussian multi-scale to improve Harris operator to detecte point feature of core iamges,and use SIFT operator to describe point feature,and then use Euclidean distance to match features.Experimental results show that using this algorithm can obtain more accurate core panoramic,and can enhance the accuracy of the core image stitching system on basis of previous,and this method has some practical value.