论文部分内容阅读
为了提高在高水平噪声下的去噪效果,提出了一种适合对受随机冲击噪声污染图像重建的两步去噪方法。利用图像的全局和局部信息来定位被噪声污染的像素,对被噪声污染像素利用基于B-Sp line插值的方法来进行重建,并且根据被污染像素周围像素的污染情况自适应地调整对其的加权因子。实验结果表明:通过B-Sp line插值来恢复被噪声污染像素的方法,对污染噪声的水平具有很好的鲁棒性,特别是在噪声水平高于50%的情况下,重建图像的峰值信噪比PSNR比传统方法平均可以提高2~3 dB。
In order to improve the denoising effect under high level of noise, a two-step denoising method suitable for image reconstruction by stochastic impact noise is proposed. The global and local information of the image is used to locate the noise-contaminated pixels. The noise-contaminated pixels are reconstructed by a method based on B-Sp line interpolation and adaptively adjusted according to the pollution of the pixels around the contaminated pixels Weighting factor. The experimental results show that the method of recovering pixels contaminated by noise by B-Sp line interpolation is robust to the level of pollution noise. Especially when the noise level is higher than 50%, the peak value of reconstructed image Noise ratio PSNR than the traditional method can be an average of 2 ~ 3 dB.