基于支持向量回归的图像复原方法研究

Image Restoration Method Based on Support Vector Regression

针对退化图像复原问题,提出了一种基于支持向量回归的退化图像复原方法.该方法利用支持向量机回归算法非线性映射能力,通过训练样本对的学习训练,在退化图像与原始清晰图像之间建立映射关系,然后对测试样本进行复原.实际图像复原实验表明,得到的复原图像在视觉上和定量分析上都获得了比较好的效果.与神经网络方法相比,支持向量机回归算法克服了神经网络的模型选择与过学习问题、局部极小问题等.

A new image restoration method is presented and investigated based on support vector regression （SVR）. The mapping relationship between degenerated image and clear image is established by train- ing support vector machine. Experimental results show that satisfactory restoration effect is obtained both in visual impression and quantitative analysis. Compared with neural network, the SVR has prominent advantages in selecting model, overcoming over-fitting and local minimum, etc.