基于噪声-纹理检测算子的图像去噪方法

利用能量泛函极小化方法对图像进行滤波时,通常用分段常数函数来近似图像,在滤除噪声的同时也丢失了许多纹理和细节信息.基于这一不足,本文提出一个噪声—纹理检测算子,利用这一算子对滤掉的信息作进一步检验,从而尽可能多的抽取出被误滤掉的纹理信息,将这些纹理信息补充回滤波后的图像中得到最终的去噪图像.实验表明,本文提出的算子对去噪后图像纹理信息的保留具有明显效果.

An Image Denoising Method Based on a Noise-Texture Operator

Denoising algorithm based on gradient dependent energy functional,modify images towards piecewise constant functions.Important information,encoded in image features like textures or certain details,is often compromised in the process of denoising.A noise-texture detect operator is proposed in this paper.The filtered information,during the process of denoising,will be checked again by using this operator,then textures and details filtered by mistake will be extracted as much as possible.After refilling these textures and details into the denoised image,the final denoised image is obtained.Experiment results show that our new method has obvious effect in preserving textures and details.