Contourlet变换针对指纹图像的重构情况研究

Research on Fingerprint Image Reconstruction Based on Contourlet Transform

Contourlet变换是一种"真正"的二维图像表示方法。它通过拉普拉斯金字塔结构来捕获奇异点,并利用方向滤波器组将分布于不同方向的奇异点合成为一个系数,用类似于轮廓段的基函数来逼近图像。它是一种多分辨的、局域的、多方向的图像表示方法。由于指纹图像含有丰富的方向性信息,所以文章主要针对指纹图像,采用Contourlet变换,在选择相同个数大系数的条件下,选择不同的滤波器、不同的级数,观察图像重构的情况,并采用峰值信噪比(PSNR)来度量重构性能。在此基础上,在相同的条件下,采用db2与Contourlet变换,分别对指纹图进行了重构,并对重构结果进行了对比,对比结果突出了Contourlet变换对图像重构的优势。

Contourlet transform is a ＂real＂ two-dimensional image representation method. The singular points are captured by the Laplacian Pyramid（LP） structure, and the singular points distributed in different directions are synthesized into a coefficient by directional filter banks, and the image is approximated by a basis function similar to the contour segment. It is a multi-resolution, local,multi-directional image representation method. Because fingerprint image contains abundant directional information, this paper, mainly aimed at fingerprint image, adopts Contourlet transform, under the condition of selecting the same number of large coefficients, choosing different filters and different series, observing the situation of image reconstruction. Peak signal-to-noise ratio（PSNR） is used to measure the reconstruction performance. On this basis, under the same conditions, DB2 and Contourlet transform are used to reconstruct fingerprint, and the reconstruction results are compared. The comparison results highlight the advantages of Contourlet transform in image reconstruction.