序列图像拼接过程中误匹配的消除  

ELIMINATION OF PSEUDO MATCHING IN IMAGE STITCHING

作　　者：李从利[1] 韩裕生[1] 冷晓艳[1] 薛模根[1] 

机构地区：[1]解放军炮兵学院信息工程教研室,安徽合肥230031

出　　处：《计算机应用与软件》2006年第11期93-95,共3页Computer Applications and Software

摘　　要：序列图像拼接过程中困难之一是当重叠区域内存在多个相似部分时,就会发生误区配而导致拼接失败。在采用图像特征区域配准算法中,基于同名区域面积差、区域重心同位、区域重心距离比及配准参数复用提出了四种误区配的消除方法。实验表明,这些方法能有效地消除误匹配,提高序列图像拼接的配准率。One of the difficulties in sequential image stitch processing is that it may be frustrated in mismatching when there are many similar segments in overlapped areas. If image characteristic areas registration algorithm was adopted, the paper put out four eliminating methods based on area homonymic proportion margin,area Centro baric parity and interval,and registration parameters repeating usage. Experiments demonstrated that this mismatching eliminating method could eliminate mismatch effectively and improve registration efficiency of sequential image stitching.

关 键 词：图像拼接 误匹配 连续帧图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]