计算机工程 ›› 2019, Vol. 45 ›› Issue (2): 207-212.doi: 10.19678/j.issn.1000-3428.0049206

• 图形图像处理 • 上一篇    下一篇

基于慢特征分析的智能拼图算法

吴娟,陈丽芳   

  1. 江南大学 数字媒体学院,江苏 无锡 214122
  • 收稿日期:2017-11-07 出版日期:2019-02-15 发布日期:2019-02-15
  • 作者简介:吴娟(1990—),女,硕士研究生,主研方向为图像处理;陈丽芳,副教授、硕士。
  • 基金项目:

    国家科技支撑计划(2015BAH54F01)。

Intelligence Puzzle Algorithm Based on Slow Feature Analysis

WU Juan,CHEN Lifang   

  1. School of Digital Media,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:2017-11-07 Online:2019-02-15 Published:2019-02-15

摘要:

现有拼图算法对背景单一、存在大量相似物的图片进行组合拼接时,不能精确分辨拼图块间的微小差异,还原的图片存在偏差。为此,提出一种智能拼图算法,通过计算相邻拼图块边缘的慢特征值选择正确的拼图块,利用贪婪算法根据拼图块的邻近关系实现图片智能拼接。实验结果表明,与MGC算法相比,该算法具有更高的拼图准确率及稳定性。

关键词: 智能拼图, 慢特征分析, MGC算法, 贪婪算法, 最小生成树

Abstract:

Existing intelligence puzzle algorithms can not restore the image well when they have a single background and a large number of similar things.Because they can not distiguish the tiny difference between puzzle block.Aiming at these problem,this paper proposes a intelligence puzzle algorithm.It selects the correct puzzle pieces by calculating the slow eigenvaluess of the edges of adjacent puzzle pieces,and then uses the greedy algorithm to realize intelligence puzzle according to the neighbor relationship of puzzle pieces.Experimental results show that the algorithm has a higher puzzle accuracy stability than the MGC algorithm.

Key words: intelligence puzzle, Slow Feature Analysis(SFA), MGC algorithm, greedy algorithm, minimal spanning tree

中图分类号: