《Image super-resolution representation via image patches based on extreme learning machine》.pdf

International Conference on Software Engineering and Computer Science (ICSECS2013) Image super-resolution representation via image patches based on extreme learning machine Qiuxi Zhu, Xiaodong Li, Weijie Mao Department of Control Science and Engineering Zhejiang University Hangzhou, China e-mail: zhuqiuxi0743@126.com Abstract—In this paper, aimed at the extensively existing problem and make it possible to increase the resolution of LR images by of slowness in mainstream image super-resolutions, an efficient more than 3 or 4 times [2]. However, most of the applied well- approach is proposed for super-resolution based on the extreme developed learning algorithms such as back propagation (BP) learning machine (ELM) for single-hidden layer feedforward and support vector machine (SVM), are still trapped by the neural networks (SLFNs). Features and issues (e.g. parameter bottleneck of slowness caused by iterative solutions and thus selections) in the application of ELM are discussed, on the basis fail to be accepted in fields that lay stress on the speed of of which a general framework for a variety of super-resolution imaging [2]. problems is proposed, and corresponding experiments are In this paper, extreme learning machine (ELM) [10] for conducted