计算机与现代化 ›› 2013, Vol. 1 ›› Issue (5): 87-89,9.doi: 10.3969/j.issn.1006-2475.2013.05.020

• 图像处理 • 上一篇    下一篇

基于注意机制的水下目标检测方法

董欣   

  1. 河海大学计算机与信息学院,江苏南京211100
  • 收稿日期:2013-02-28 修回日期:1900-01-01 出版日期:2013-05-28 发布日期:2013-05-28

Underwater Target Detection Method Based on Attention Mechanism

DONG Xin   

  1. College of Computer and Information, Hohai University, Nanjing 211100, China
  • Received:2013-02-28 Revised:1900-01-01 Online:2013-05-28 Published:2013-05-28

摘要: 针对水下光学图像清晰度低和物体颜色失真造成水下目标检测困难的问题,本文提出一种基于注意机制的水下目标检测方法。在最具代表性的视觉注意模型(Itti模型)基础上,在特征提取阶段,引入带色彩恢复的多尺度Retinex算法对图像的颜色信息进行恢复和增强。针对具体的水下环境,通过实验构造带彩色恢复的多尺度Retinex算法中的增强因子,从而构建一种新的适用于水下环境的目标检测方法。实验结果表明,本文的算法能有效地检测出目标且能得到较完整的目标区域。

关键词: 水下图像, 目标检测, Retinex, 注意机制

Abstract: In view of the low underwater optical image resolution and the color distortion in underwater target detection, this paper proposes an underwater target detection method based on attention mechanism. According to the most representative visual attention model (Itti model), in the feature extraction stage, this paper introduces the multiscale Retinex algorithm to restore and enhance the images’ color information. Through the experiment to get a suitable enhancement factor of the multiscale Retinex algorithm, this paper constructs a new suitable target detection method for the specific underwater environment. Experimental results show that the algorithm can effectively detect the target and complete target areas can be obtained.

Key words: underwater images, target detection, Retinex, attention mechanism

中图分类号: