Abstract—As the rapid development of the multi-media technology, more and more digital devices are used to take pictures. However, sometimes the images may be degraded due to disoperation, the environment, and system and so on. The main task for image recovery is combining the degraded images and some priori information to acquire the best estimation of the original image by using some kind of restoration algorithm. In this paper, a new image restoration algorithm based on ELM Neural Network and some edge information was proposed, aiming at the fuzzy motion images. As the experiment shows that comparing traditional BP Neural Network, the new restoration algorithm based on ELM Neural Network is simpler, faster, easier to implement, what’s more, it’s more suitable for real-time processing.

Index Terms—Image Restoration; Neural Network; ELM algorithm; Fuzzy Images; Best Estimation

Cite: Jianhong Zhou, Yong Feng, and Deyuan Tao, "Research on ELM-based Image Restoration Algorithm," Journal of Communications, vol. 8, no. 8, pp. 529-534, 2013. doi: 10.12720/jcm.8.8.529-534