文章摘要
唐艳,孙刘杰,王勇.基于SIFT变换和RBF神经网络的彩色全景图拼接算法[J].包装工程,2018,39(21):216-221.
TANG Yan,SUN Liu-jie,WANG Yong.Color Panorama Mosaic Algorithm Based on SIFT Transform and RBF Neural Network.Packaging Engineering,2018,39(21):216-221.
基于SIFT变换和RBF神经网络的彩色全景图拼接算法
Color Panorama Mosaic Algorithm Based on SIFT Transform and RBF Neural Network
投稿时间:2018-06-29  修订日期:2018-11-10
DOI:10.19554/j.cnki.1001-3563.2018.21.037
中文关键词: 图像处理  图像拼接  尺度不变特征变换  RBF神经网络
英文关键词: image processing  image stitching  scale invariant feature transform  RBF neural network
基金项目:
作者单位
唐艳 上海理工大学上海 200093 
孙刘杰 上海理工大学上海 200093 
王勇 上海理工大学上海 200093 
摘要点击次数: 51
全文下载次数: 65
中文摘要:
      目的 为了复原存在平移、色彩差异、旋转、形变等问题的全景图,提出一种结合SIFT(尺度不变特征变换)和RBF神经网络的彩色全景图拼接算法。方法 通过SIFT算法匹配出两子图中对应的特征点,利用仿射变换解决图像间的旋转和形变问题,采用RBF神经网络纠正子图的色彩差异,最后利用权值矩阵融合技术实现重叠区域的融合。结果 文中算法在拼接效果上优于其他算法,其拼接效果DoEM值为0.902,图像重叠区域过度平滑,有效地避免了融合区域的亮度块或亮度线。结论 该算法效果好,可解决全景图复原过程中多方面的难题。
英文摘要:
      In order to restore the panorama with the problems of translation, color difference, rotation and deformation, a color panorama stitching algorithm combining SIFT (scale invariant feature transform) and RBF neural network is proposed. The SIFT algorithm was used to match the corresponding feature points in the two subgraphs, and the affine transformation was used to solve the rotation and deformation problems between images. The RBF neural network was used to correct the color difference of the subgraphs. Finally, the weight matrix fusion technique was used to realize the fusion of overlap areas. The proposed algorithm was superior to other algorithms in the stitching effect. The DoEM value of the stitching effect was 0.902, and the image overlap area was excessively smooth, effectively avoiding the luminance block or the luminance line of the merged area. The proposed algorithm has good effect and can solve many problems in the panorama restoration process.
查看全文   查看/发表评论  下载PDF阅读器
关闭

关于我们 | 联系我们 | 投诉建议 | 隐私保护 | 用户协议

您是第4978612位访问者    渝ICP备10003521号-2

版权所有:《包装工程》编辑部 2014 All Rights Reserved

邮编: 电话:023-68792294传真: Email:

    

渝公网安备 50010702501716号