基于第二代Curvelet变换的自适应图像增强

Adaptive image enhancement algorithm based on second generation curvelet transform

Curvelet是继小波和Ridgelet之后一种新的图像多尺度表示方法,Curvelet具有多尺度,多方向的特性,属于高度各向异性的变换。第二代Curvelet变换克服了第一代Curvelet变换的高数据冗余度问题,特别是基于"Wrapping"方式的第二代离散Curvelet算法,不仅运算快速、几何真实,而且快速可逆。因此,将第二代Curvelet变换用于图像增强,并通过自适应地确定Curvelet分解子带的噪声水平,实现了一种自适应图像增强方法。实验结果表明,同基于小波变换的图像增强方法相比,该方法具有明显的优势。

Curvelet,as a new multiscale analysis algorithm which is an extension and latest development of Wavelet and Ridgelet,is a kind of muhiscale,muhi-directional and anisotropic transform.The second generation Curvelet,espeeially the “wrap- ping” version of the second generation discrete Curvelet algorithm,can effectively reduce the data redundancy of the first generation Curvelet.Since the second generation eurvelet has good operate speed and is geometry faithful,based on theory of image enhancement using it,this paper proposes an adaptive image enhancement algorithm through adaptively confirming noise level of Curvelet sub-bands.The experiments show that the new method has obvious advantage compared with the image enhancement based on wavelet.