摘要
提出一种对图像DCT(离散余弦变换)系数使用稀疏表达的彩色图像增强算法。该算法改进了传统算法在边缘保留和噪声抑制方面的一些不足,例如增强高频成分中的纹理信息的同时抑制噪声。传统的图像增强算法往往认为图像的高频成分主要为噪声区域,在这样的前提下高频成分中的纹理信息和噪声区域就很难加以区分。所提出的算法主要依据图像上不同区域DCT系数的不同分布特征识别图像的噪声和非噪声频率成分,进行噪声抑制和边缘增强,并且使用Kaiser加窗处理来解决因保留低频成分产生的块效应。实验结果表明,与传统图像增强算法相比,所提出的算法不仅抑制了图像噪声,而且增强了图像的纹理和边缘,改善了图像的视觉效果。
We propose a colour image enhancement method which uses the sparse representation on DCT coefficients of the image. The method overcomes the shortcomings of traditional algorithms in edge preservation and noise suppression, for example, it enhances the texture information in high frequency component and suppresses noise at the same time. Traditional image enhancement algorithms usually consider that the noise is the main area in high frequency component, therefore it is hard to distinguish the texture information and the noise area in high frequency component with such premise. The algorithm proposed in the paper identifies the noise and non-noise area in the image and carries out noise suppression and edge enhancement mainly relying on the different distribution features of DCT coefficients in different areas of the image. Besides, the Kaiser window function is used to remove the blocky effect caused by low frequency component preservation. Experimental results demonstrate that compared with traditional image enhancement algorithm ,. the algorithm presented in this paper enhances the texture and edges of the images while suppressing the noise in the image, and improves the visual effect of the image.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第6期236-239,263,共5页
Computer Applications and Software
关键词
稀疏表示
图像增强
离散余弦变换
噪声抑制
Sparse representation Image enhancement Discrete cosine transform （DCT） Noise suppression
