Real-Time Neuromorphic Systems For Image Processing - Springer

Recently, a new discipline has come up which challenges classical approaches to computer vision research and engineering, the so-called neuromorphic engineering. This new research paradigm develops engineering systems with unparalleled robustness and on-line adaptability. Such systems are based on algorithms ...
52KB Sizes 0 Downloads 77 Views
JOURNAL OF REAL-TIME IMAGE PROCESSING SPECIAL ISSUE ON REAL-TIME NEUROMORPHIC SYSTEMS FOR IMAGE PROCESSING 1. Introduction Recently, a new discipline has come up which challenges classical approaches to computer vision research and engineering, the so-called neuromorphic engineering. This new research paradigm develops engineering systems with unparalleled robustness and on-line adaptability. Such systems are based on algorithms and techniques closely replicating and inspired from the principles of information processing in biological nervous systems. The robustness of the human visual system in almost any visual situation is enviable, performing enormous calculation tasks continuously, robustly, and effortlessly. Their applicability to real-time computer vision challenges is extended, and includes prosthesis systems, robust human-computer interfaces, smart sensors, implanted electronic devices, autonomous visually-guided robotics systems. Neuromorphic engineering fills the gap between, on the one hand, computational neuroscience, and, on the other hand, traditional engineering. Computational neuroscience has yielded very useful models and theories of brain function, although often these have been restricted to simplified conditions, stimuli and tasks, to allow direct comparison with simple empirical measurements on biological systems. Also, computational neuroscience models typically are concerned with testing one given hypothesis, and thus are not intended to solve real-world problems, but they advance the understanding of brain function through hypothesis testing. Thus, those models often tend to not scale-up to more task conditions, environment or complex stimuli. Traditional Engineering on the contrary -specifically Image Processing- has focused on developing systems that can solve actual real-world problems; however, because general problems such as recognizing objects in a digital image or driving a vehicle from one city to another are incredibly complex, image understanding solutions have been also explicitly restricted to simplified environments and tasks (i.e drone tracking in a controlled environment with advance knowledge or specific pattern recognition and classification before applying a broader problem of general scenario). Since the nature of those algorithms is complex, mostly it is needed specific parallel paradigms to achieve the performance, power, and flexibility to support neuromorphic applications needing many integration of heterogeneous architectures. The goal of this special issue is to bring together researchers and practitioners working in the area of Real-Time Bioinspired and Neuromorphic Systems for Image Processing accelerating these heavy algorithms using GPUS, FPGAs, VLSI systems, DSPs,

Multicores among other platforms. We are soliciting original contributions, which address a wide range of theoretical and practical issues related to high-performance motion estimation image processing including, but not limited to: * Real-Time Vision Neuromorphic Models * Real-Time Vision based Human-Computer Interface systems * Real-Time Vision Prosthesis, Artificial Retinas, Bionic Eyes * Real-Time Bioinspired and Psicoinspired systems * Real-Time Biometrics and Bioinspired Security Systems * Deep Convolutional Neural Networks for image processing * Energy-Efficiency aspects * Specific Architectures * Metrics for Bioinspired Systems on Real-Time Image Processing * Real-World Applications 2. Paper Submission Submitted manuscripts must be unpublished and may not be under consideration elsewhere. Manuscripts should conform to the standard guidelines of the Journal of Real-Time Image Processing, (http://www.springer.com/computer/image+processing/journal/11554). All potential authors are requested to volunteer as reviewers in the peer-review process for manuscripts submitted for this special issue. Prospective authors should submit an electronic copy of their manuscript through the Editorial Manager (https://www.editorialmanager.com/jrtip). Papers should be marked as “SI: Real-Time Neuromorphic


104 Views
73 Views