计算机工程与应用 ›› 2013, Vol. 49 ›› Issue (20): 188-192.

• 图形图像处理 • 上一篇    下一篇

多摄像机图像拼接视觉归一化技术研究

李洋洋1,赵  刚1,2,刘  利3,刘  钊3   

  1. 1.华中师范大学 国家数字化学习工程技术研究中心,武汉 430079
    2.华中师范大学 信息与新闻传播学院,武汉 430079
    3.北京市公安局 科技信息化部,北京 100740
  • 出版日期:2013-10-15 发布日期:2013-10-30

Research on visual normalization of multi-camera image stitching

LI Yangyang1, ZHAO Gang1,2, LIU Li3, LIU Zhao3   

  1. 1.National Engineering Research Center for E-Learning, Central China Normal University, Wuhan 430079, China
    2.College of Information Technology & Journalism and Communications, Central China Normal University, Wuhan 430079, China
    3.Department of Science and Technology Information, Beijing Municipal Public Security Bureau, Beijing 100740, China
  • Online:2013-10-15 Published:2013-10-30

摘要: 视觉归一化是多视点图像拼接领域的一个关键技术,在对大量图像处理算法研究的基础上,提出了一种针对多摄像机图像拼接的视觉归一化处理方法。该方法主要包括图像颜色校正和图像边缘融合两个模块;在图像颜色校正模块中,引入了图像区域划分策略和自适应颜色调节因子,使不同的像素点都有不同的颜色调节因子,并充分利用相邻图像间的颜色关联性对目标图像的颜色进行自适应校正;在图像边缘融合模块中,利用反映射矩阵计算出拼接图像的重叠区域,利用自适应边缘融合因子对重叠区域进行边缘融合处理。实验结果表明,该方法能够较好地减少甚至消除拼接图像间的视觉差异,较好地改善了图像拼接的视觉效果。

关键词: 视觉归一化, 颜色校正, 边缘融合, 图像拼接

Abstract: Visual normalization is a key technology in multi-view image stitching . In this paper, based on the research of image processing algorithms, a novel method of visual normalization is presented for multi-camera image stitching, which mainly includes image color correction and image edge fusion. In processing of color correction, image region and self-adaptive strategy is used to make that different pixels have different blend factors, and the color relevance between adjacent images is also considered. In processing of edge fusion, the reflect matrix is used first to calculate the overlapping of two images, and then fill the overlapping by different colors resulted from two images using different blend factors in each pixel. The experimental result shows the efficiency of the method presented by this paper, which can effectively improve the visual effects in multi-view image stitching.

Key words: visual normalization, color correction, edge fusion, image stitching