基于双树复小波变换的图像增强方法

An Image Enhancement Method Based on Dual-Tree Complex Wavelet Transform

在图像增强处理中,传统的图像增强方法例如直方图均衡、小波系数增强等等,虽然取得了较好的图像增强效果,但在增强图像的同时也将噪声放大了。本文针对这一问题,提出了一种基于二维双树复小波变换的图像增强方法,因其具有良好的多方向性信息捕捉能力,克服了传统二维小波变换缺乏方向性的缺点。增强算法将系数分为强边缘、弱边缘和噪声点三类,通过增强函数进行分别处理,达到有效抑制噪声的同时增强图像的边缘信息。文中以PSNR为评价准则,讨论了参数对增强函数的影响,来优化参数选取。实验结果表明,本文所提出的方法得到了较好的图像增强效果。

In traditional image enhancement methods, such as histogram equalization, wavelet transform, etc, these methods achieve good performance, however, there still exist some drawbacks. The noise is amplified during the image enhancement processing. In this paper, an image enhancement method based on the 2D dual-tree wavelet transform is proposed, aiming at the exsiting problem in traditional image enhancement methods. The 2D dual tree complex wavelet transform can describe the directional information of images better. The enhancement algorithm divides coefficients into three classes： strong, weak edge and noise point. The enhancement function will process the coefficients of each class respectively. The enhancement method adopted in this paper can suppress noise effectively. In this paper, the effect of parameters on the enhancement function is discussed taking PSNR as the evaluation standard, in order to optimize parameter selection. The experimental results show that the proposed method receives better performance.