摘要
为了对电力设备故障进行更为有效和直观的诊断,提出了一种基于灰度冗余和SURF算法的红外和可见光图像的配准方法。该算法对红外图像和可见光图像进行基于灰度冗余的处理,利用SURF方法分别找到两幅图像的特征点并求出其描述子,通过RANSAC(随机采样一致性)对这些特征点进行精确配对,通过最小二乘法求得仿射变换参数并实现两幅图像的配准。实验表明,该方法与其他配准方法相比,具有速度快,鲁棒性高等优点,具有较高的实用价值。
In order to carry out more effective and intuitive fault diagnosis for electrical equipment,this paper presents a registration algorithm based on redundancy of gray level and SURF（speeded up robust features）for the infrared image and visible image.Firstly,the infrared images and visible images are processed based on gray-scale redundancy.Then the interest points of two images are detected respectively based on SURF algorithm and their descriptors are calculated.A RANSAC technique is applied to make accurate matching,and then affine transformation parameters are obtained by a least-squares solution.Finally,the registration between the infrared image and visible image is achieved using these parameters.Experimental results show that the method has stronger robustness and faster speed compared with other registration algorithms,and has high practical value.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2011年第11期111-115,123,共6页
Power System Protection and Control
基金
中央高校基本科研业务费青年教师科研项目(2009JS100)
