Special issue on deep learning for emerging embedded real-time image and video processing systems

One of the main aims of the multimedia as related to image and video processing is to enable real-time image super resolution or a visually pleasing high-resolution image based on low-resolution image sequences. High resolution images are composed of higher pixel density with fine and more precise details as compared with low-resolution images or video. Many related applications, such as video surveillance, ultra-high definition TV, low-resolution face recognition, and remote image sensing are based on super-resolution techniques. These techniques have attracted high interest from both academia and industry, and currently is an active area of research in image and video processing. Previously, conventional machine learning techniques, such as supervised and unsupervised learning, reinforcement learning, Bayes classifier, K-means clustering, random forests, and decision trees, etc., have been utilized. Recently, the rapid advancements in deep learning or deep neural networks have shown a promising performance for high resolution scenarios. There remain many research issues regarding the high-resolution aspect. The objective of this special issue is to the application of deep learning for real-time super resolution image and video processing, including new objective functions, new architectures, large scale images, depth images, data acquisition, feature representation, knowledge understanding, and semantic modeling, types of corruption, and new applications. There still exists a gap between extracting representations (or knowledge) from high resolution image and video data and their practical demands