Efficient video mosaicking for partial high-resolution image adaptive to region

Methods for precisely digitizing a selected region of a still planar image by using a sequential image frame obtained by a conventional video camera are attractive. However, these methods have disadvantages, including accumulation of errors in matching frame positions, and high processing cost. This paper analyzes conventional methods and then proposes a new method to overcome these disadvantages. In the proposed method, many image frames are divided into groups of similar images, and a stable image (which is unlikely to cause position errors) is selected from each group, after which the stable images are combined. The proposed method produces high‐resolution images at lower processing cost, including hierarchical image‐mosaics. This paper also shows quantitatively, by experiments and analyses, that the proposed method is a highly effective video mosaicking system. © 2005 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 88(6): 34–44, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20189