摘要
基于学习的超分辨率复原算法是目前图像复原领域最具有潜力的方法之一。针对现有算法遍历搜索样本库,运算复杂度高且存在误匹配现象等问题,本文提出了一种新的基于预分类学习的超分辨率复原算法。算法根据简单的纹理特征对样本库进行预分类,分成若干子样本库,然后在子样本库中进行像素级精确匹配搜索。预分类过程的引入,既有效降低了精确匹配的复杂度,又因有效利用了样本的纹理特征,提高了子样本库内容的相关性,减少了误匹配。实验表明,本文提出的算法能有效提高算法结果的复原质量和运行速度。
Learning-based image super-resolution is one of the most promising approaches to solve the image super-resolution problem. A novel pre-classified learning based image superresolution algorithm is proposed to reduce the complexity of full searching and to avoid mismatching. A texture-based pre-classified process is used to select a subset of samples. Then, the best-matching samples are searched among the selected subsets. In the proposed algorithm, the complexity of the searching process is effectively reduced by the texture-based preclassified process. Furthermore, using the texture features, the mismatching probability is reduced. Experimental results show that both the visual quality and the run-time are improved.
出处
《数据采集与处理》
CSCD
北大核心
2009年第4期514-518,共5页
Journal of Data Acquisition and Processing
基金
国家自然科学基金(60431020
60772069)资助项目
北京市自然科学基金(4062006)资助项目
关键词
纹理特征
学习
样本库
超分辨率
texture features
learning
training set
super-resolution
