許陳鑑王偉彥Hsu, Chen-ChienWang, Wei-Yen黃紹慈Huang, Shao-Tzu2019-09-032022-08-312019-09-032017http://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22G060475017H%22.&%22.id.&http://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/95672匹配兩張影像之高維度特徵點，是在電腦視覺領域的眾多應用中花費大量計算資源的一環。雖然透過降低特徵點維度的手段得以抑制計算量，但是會因而犧牲了匹配的精準性。因此，本文提出一改良式的影像匹配演算法，運用K-means分群的特性，不僅可以有效地降低匹配所需的運算時間，同時也保有了一定程度的精準性。實驗結果顯示，與參考的文獻方法相較，本文所提出的方法在精準度上較具優勢。另外，為提升演算法的執行效能，本文也利用FPGA實現所提出之影像匹配演算法，藉由管線式的硬體設計架構，進一步提升影像匹配的速度。Matching high dimensional features between images is computationally expensive for exhaustive search approaches in computer vision. Although the dimension of the feature can be degraded by simplifying the prior knowledge of homography, matching accuracy may degrade as a result. In this thesis, we present a feature matching method based on K-means algorithm, which combines with L1-norm based pyramid structure that reduces the matching cost to match the features between images instead of using a simplified geometric assumption. Experimental results show that the proposed method outperforms the previous linear exhaustive search approaches in terms of the inlier ratio of matched pairs. We also implement the proposed approach on FPGA using a structured pipeline design to further improve the execution efficiency of the proposed matching algorithm.線性搜尋法K-means分群特徵點匹配feature matchingK-means clusteringlinear exhaustive search