摘要
锯齿波、边缘模糊是影响图像质量的重要因素,为了有效提高图像质量,提出了一种优化支持向量机图像插值方法.先将遗传算法应用到支持向量机的参数寻优,将使用最优参数的支持向量机结合图像相关性对图像插值.算法根据图像相关性选择适当的相邻点作为输入模式训练支持向量机,用训练好的支持向量机及输入模式估计出待插值点的像素值.仿真结果表明,与已有算法相比,该算法获得图像的RPSN值、NMSE值、MSE等指标均有明显改善,且视觉效果有显著提高.
Saw tooth wave and blurring are the important factors of affecting image quality. In order to improve the image quality, a new genetic algorithm parameters selection based image interpolation scheme using support vector machine was proposed. The genetic algorithm was applied to support vector machine parameters optimization first. The selected optimal param- eters combination the image correlation applied to the image interpolation. The neighbor training sample models were selected with considering correlation and the unknown gray val- ue was estimated by the trained SVM and input pattern. Simulation results showed that the proposed scheme produced visually pleasing NMSE and MSE than other well-known image images and obtains higher RPSN and smaller interpolation algorithms.
出处
《哈尔滨商业大学学报（自然科学版）》
CAS
2013年第2期212-217,共6页
Journal of Harbin University of Commerce：Natural Sciences Edition
基金
安徽省杰出青年基金(1108085J03)
安徽理工大学博士科学研究基金(11223)
安徽理工大学青年教师科学研究基金(12257)
淮南市指导性科技计划项目(2011B31)
关键词
图像插值
支持向量机
参数寻优
遗传算法
支持向量回归机
image interpolation
support vector machine （SVM）
parameters optimization
ge-netic algorithm （GA）
support vector regression （SVR）
