基于多特征结合与加权支持向量机的图像去噪方法

在基于支持向量机(SVM)的图像去噪方法的基础上，提出了一种基于多特征结合与加权SVM的图像去噪方法。首先，根据图像中相邻像素的相关性及椒盐噪声的特点，提取含噪图像中的多种特征；然后，利用针对不平衡数据集所改进的加权SVM分类器，识别出含噪图像中的噪声点，再利用支持向量回归机(SVR)对噪声点的原始灰度值进行回归预测；最后，重构图像以达到去噪的目的。实验结果表明，该方法能提高SVM分类器对噪声点的识别率，改善分类器的性能，并能在去噪的同时较好地保留图像的边缘信息，获得较高的峰值信噪比(PSNR)。

Image de-noising method based on multi-feature combination and weighted support vector machine

The authors put forward an image de-noising method by combining multiple features with weighted Support Vector Machine (SVM) based on the image de-noising by using SVM. Firstly, according to the adjacent pixels correlation in the image and the characteristics of salt-pepper noises, multiple features were extracted from noisy image. Then the noise points in the noisy image were detected by using weighted SVM classifier which improved on imbalanced dataset, then Support Vector Regression (SVR) was used to forecast the gray value of noise points, finally the image was reconstructed so as to remove noise points. The experimental results show that the proposed method can improve the capability of classifier and the recognition rate of noise points. Moreover, it retains the information of image edge when removing noise points, and obtains higher Peak Signal-to-Noise Ratio (PSNR).