改进的SIFT算法在图像特征点匹配中的应用

为了提高图像拼接过程中常用的SIFT(尺度不变特征)算法的特征点匹配准确率,减少误匹配特征点的数量,为后续的图像拼接提供准确的依据,通过将SIFT算法和RANSAC(随机抽样一致性)算法相结合,提出了一种提高SIFT算法匹配准确率的算法。在利用SIFT算法对目标图像进行特征提取以及特征点匹配后,再由RANSAC算法利用迭代方式估算出一个合理的数据模型,剔除掉不符合该模型的错误匹配点。最后利用该算法得到的匹配特征点进行图像拼接,拼接后的结果表明该算法准确、有效。

Application of Improved SIFT Algorithm in Image Feature Point Matching

In order to improve the accuracy of feature point matching of SIFT(Scale Invariant Feature Transform)algo-rithm in the process of image mosaicking,reduce the number of mismatched feature points and provide accurate evi-dence for subsequent image mosaicking,an algorithm for improving the matching accuracy of SIFT algorithm was pro-posed by combining the SIFT algorithm with the RANSAC(Random Sample Consensus)algorithm. After the SIFT algorithm is used to extract the target image and match the feature points,a reasonable data model is estimated by iter-ative method with RANSAC algorithm;and then the error matching points that do not conform to the model are elimi-nated;and finally the matching feature points obtained by the algorithm are used for image mosaicking. The result shows that the algorithm is practicable and effective.