Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (10): 1-3.

• 博士论坛 •     Next Articles

A New Sonar Image Denoising Method Based on Circular Sampling FRIT

Zheng-guo SHANG Chun-hui ZHAO Yan SUN   

  • Received:2006-11-29 Revised:1900-01-01 Online:2007-04-01 Published:2007-04-01
  • Contact: Zheng-guo SHANG

基于FRIT循环抽样声纳图像去噪新方法

尚政国 赵春晖 孙岩   

  1. 哈尔滨工程大学 哈尔滨工程大学 哈尔滨工程大学
  • 通讯作者: 尚政国

Abstract: The sonar image has the complex background, bad contras and deteriorate edges. It is difficult to detect the edges of the object. The Wavelet transform used in the sonar image denoising could not do well in the edge keeping. The finite Ridgelet transform (FRIT) is a new method of high dimension signal processing. It conquers the defects of the wavelet, which could not deal with the singularity in high dimension well. The circular sampling FRIT is first used in the sonar image denoising technique, and makes high improvement to heighten the Signal Noise Ratio. Compared with other methods, it obtains the good result in the sonar image denoising and edge keeping.

Key words: Sonar image, Image processing, Ridgelet Transform, Wavelet Transform, Circular Sampling

摘要: 声纳图像背景复杂,对比度差,边缘恶化,不易判读图像边缘。对声纳图像执行小波变换能够有效去除噪声,但是由于小波的局限性,其对图像边缘的保持效果不佳。有限Ridgelet变换 (FRIT)能够有效克服小波变换在处理高维信号时的不足,是一种有效处理二维奇异性信号的新方法。将FRIT处理技术应用在水下声纳图像去噪技术中,基于该方法提出循环抽样FRIT去噪算法,提高了处理结果的信噪比及边缘保持效果。在实验数据比较中,此改进算法好于其他经典方法。

关键词: 声纳图像, 图像处理, Ridgelet变换, 小波变换, 循环抽样