应用图像纹理连续性的4f系统图像非下采样轮廓波变换域降噪

4f系统同时具有噪声和低通特性,为实现对其输出图像降噪的同时保护图像细节,提出一种结合图像纹理连续性的非下采样轮廓波变换域去噪方法。在传统NSCT域硬阈值去噪方法基础上,首先,对高频子带图像用一个衰减的阈值去除小幅值的噪声点,以更多地保护图像细节并凸显剩下的大幅值噪声点的孤立性。然后,利用图像细节纹理连续性分布和剩下的大幅值噪声点孤立分布的区别分离余下的图像细节和噪声点,进一步实现去噪的同时更好地保护图像细节的目的。实验结果表明,与传统方法相比,该方法峰值信噪比(PSNR)提高了0.5~1dB,结构相似度指数(SSIM)提高了3%~5%,因此能更好地保护图像细节,视觉边缘效果更清晰。

4f system image denoising in NSCT domain based on continuity of image texture

4f system has the characters of noise and low-pass.In order to remove the image noise and to protect the image texture of its outputs,an image denoising method based on Nonsubsampled Contourlet Transform(NSCT) is proposed combined with the continuity of image texture. Based on the classic hard threshold denoise method in NSCT domain, it adopts a smaller threshold to remove the small amplitude noise to protect the image detail better and protrude the remanent big amplitude noise, then uses the continuity of ...