目的 分析 SIFT算法特征点描述子生成比较缓慢和匹配过程距离比阈值无法自适应调节的问题， 探索一种改进SIFT的图像快速自适应匹配算法。 方法 通过简化特征点描述子的生成过程， 提高算法效率， 并通过自适应地调节距离比阈值参数， 提高算法的鲁棒性。 结果 与原算法相比， 图像快速自适应匹配算法的匹配点数减少， 匹配准确度高， 计算时间缩短。 结论 该算法在匹配效率和准确度方面具有明显优势， 能够将计算机视觉用于工业无损检测、印品图像检测、印刷网点图像检测和包装图像检测中，具有较好的推广价值。

Objective To analyze the problems of very slow generation of feature point descriptors and incapacity of self adaption and adjustment of the distance ration threshold during the matching process in original SIFT algorithm, and to propose a fast adaptive image matching algorithm based on improved SIFT. Methods The efficiency of the algorithm was improved through simplifying the feature point descriptor generation process, and the robustness of the algorithm was improved through adaptive adjustment of the distance ration threshold parameters. Results The test results showed that the algorithm matched more accurately with smaller amount of calculation, though the matched points became fewer. Conclusion With the advantages of efficiency and accuracy, the proposed algorithm was suitable for computer vision in industrial nondestructive detection, printing image detection, printing dot image detection and packaging image detection,with good promotional value.