基于L-曲率流滤波器的图像降噪算法

Image denoising algorithm based on L-curvature flow filter

提出了L-曲率流滤波器的图像降噪(滤波)算法,该方法按图像信噪比大小分高、中、低3类,分别由L滤波器降噪、多级L滤波器降噪以及多次迭代的组合滤波器降噪,并进行了实验研究。结果表明:该算法与均值和中值滤波器相比,输入图像信噪比越低,滤波效果越明显。当输入图像为低信噪比时,对于受高斯噪声污染的图像,该算法滤波比均值滤波平均提高2.98 dB;对于受脉冲噪声污染的图像,该算法滤波比中值滤波平均提高11.09 dB,说明该算法对降低不同种类和不同信噪比的图像噪声有较强的适应性。

Denoising algorithm of L-curvature flow filter was presented. The image noise was obviously removed according to SNR level in terms of the algorithm. L filter, multistage L filter, and the filter combined L filter with curvature flow filter through many iterations, could filter image noise of higher, middle, and lower SNR level respectively. Experiment results show that the lower the input image SNR is, the better the performance of developed algorithm is comparing with the average filter and mean filter. When input image SNR level is low, output image SNR of the algorithm is about 2.98 dB higher than average filter for images with Gaussian noise, and 11.09 dB higher than mean filter for images with impulse noise. The method is very efficient to decrease the image noise of different kinds of SNR and intensities.