新型三维形貌测量系统SIFT算法改进研究

摘要

The tri-linear CCDs photogrammetry using in surveying was applied to large scale 3-D shape measurement. To pick out the homologous image points from tri-linear CCDs images,SIFT algorithm was used final-ly. In order to against deficiencies of traditional SIFT,such as time consuming,a new improved SIFT algorithm was proposed. A searching strategy and threshold based on new system were given to meet“real-time”,verified by the experiments. The experiments show that the improved SIFT can cut down the time-consuming and increase the num-ber of matching point-pair and performance of algorithm.%将测绘领域的三线阵摄影测量原理应用于运动物体三维形貌测量，提出一种基于三线阵CCD的新型三维形貌测量方法。针对新型测量系统获取的三线阵CCD影像同名点匹配问题，研究了常用的影像匹配算法，并选用SIFT匹配算法。由于传统SIFT匹配算法处理大尺寸图像耗时长、提取匹配点数少等不足，对算法进行优化，并提出基于新型大尺寸三维形貌测量系统的匹配搜索策略，确定最优匹配阈值，最后通过对比实验进行验证。实验表明，改进的算法能够解决匹配影像视角变化等问题，缩短算法处理时间，增加匹配点数，提高算法性能。