基于BRISK特征匹配航拍图像自动配准方法

Aerial Video Image Automatic Registration Method Based on Brisk Feature Matching

摘要:

针对航拍图像存在平移、旋转、尺度和亮度变化的问题，提出了一种新的基于二进制鲁棒不变尺度特征(BRISK)的快速图像自动配准算法。首先提取BRISK特征点，其次采用引导互匹配策略得到初始匹配点对，然后采用预检测的随机抽样一致(RANSAC)算法剔除误匹配点对，最后用最小二乘法求出帧间的变换模型参数，采用双线性插值得到配准后的图像。实验结果表明，该算法不但满足配准精度的要求，而且运行速度远远快于SURF算法,具有一定的理论和应用价值。

Abstract:

Aiming at the existing problems of translation, rotation, scale and illumination variations of aerial image, proposed is a new fast automatic image registration algorithm based on Binary Robust Invariant Scalable Keypoints(BRISK) fearture matching. The method extracts BRISK feature points first，and uses the guide mutual matching strategy to get the initial matching points, then uses the pre-detection of random sample consensus (RANSAC) algorithm to eliminate the false matching points. Finally, the affine transformation model parameters are calculated by the least square method, and the registration image is obtained by using the bilinear interpolation. Experimental results show that the proposed algorithm not only meets the requirements of registration accuracy, but also runs faster than the SURF algorithm, which has a certain theoretical and application value.