作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2013, Vol. 39 ›› Issue (8): 227-230. doi: 10.3969/j.issn.1000-3428.2013.08.049

• 人工智能及识别技术 • 上一篇    下一篇

运动图像序列中的斑点检测算法

张然然1a,庞海波2,张 怡1b   

  1. (1. 天津大学 a. 计算机科学与技术学院;b. 软件学院,天津 300072;2. 上海大学计算机工程与科学学院,上海 200072)
  • 收稿日期:2012-02-29 出版日期:2013-08-15 发布日期:2013-08-13
  • 作者简介:张然然(1980-),女,博士研究生,主研方向:图像与视频处理;庞海波、张 怡,博士研究生
  • 基金资助:
    上海市教委重点学科建设基金资助项目“可视媒体修复与检索关键技术研究”(J050103)

Blotch Detection Algorithm in Motion Image Sequence

ZHANG Ran-ran   1a, PANG Hai-bo   2, ZHANG Yi   1b   

  1. (1a. School of Computer Science and Technology; 1b. School of Software, Tianjin University, Tianjin 300072, China; 2. School of Computer Engineering and Science, Shanghai University, Shanghai 200072, China)
  • Received:2012-02-29 Online:2013-08-15 Published:2013-08-13

摘要: 传统运动图像斑点检测算法通过降低阈值来提高正确检测率,但较高的错误检测率直接影响其后的斑点修复结果。为最大化正确检测率与错误检测率之间的比值,在SROD检测算法的基础上提出改进的斑点检测算法。引入自适应多阈值的概念,对同一帧图像中的不同斑点进行自动收敛,找到合适的阈值,采用纹理匹配方法减小斑点区域中由于运动矢量估计导致的偏差。对实际破损以及人工添加斑点的图像序列进行性能评估,结果表明,改进算法能在得到较高正确检测率的同时降低错误检测率。

关键词: 斑点检测, 自适应多阈值, 纹理匹配, 迭代算法, 视频修复, 运动矢量估计

Abstract: Traditional blotch detection algorithms obtain higher correct detection rates by reducing the value of threshold. However, the corresponding higher number of false alarms affects the following correction results directly. To maximize the ratio between correct detections and false alarms, this paper proposes an improved blotch detector based on simplified rank-ordered difference. The improved detector can achieve the most appropriate threshold to the different blotch in a certain frame by introducing dual-steps adaptive multi-threshold. It uses employment of texture matching to avoid the possible deviation caused by motion vector estimation in the regions with blotches. Performance evaluation is taken to the image sequences with both real blotches and artificially corrupted ones separately. Experimental results indicate higher correct detection rates and fewer false alarms simultaneously.

Key words: blotch detection, adaptive multi-threshold, texture matching, iterative algorithm, video restoration, motion vector estimation

中图分类号: