Principles of Organization of the Human Eye Retina and Their Use in Computer Vision Systems
- PDF / 339,868 Bytes
- 13 Pages / 594 x 792 pts Page_size
- 67 Downloads / 212 Views
CYBERNETICS PRINCIPLES OF ORGANIZATION OF THE HUMAN EYE RETINA AND THEIR USE IN COMPUTER VISION SYSTEMS V. P. Boyun,1† L. O. Voznenko,1‡ and I. F. Malkush1††
UDC 004.89; 004.93
Abstract. This paper provides a summary of principles of organization of the human eye retina. The following principles are considered: locality during interacting neurons, ring organization of receptive fields with on- and off-centers (center-surround organization), specialization of neuron layers, organization of feedbacks, adaptation to lighting and contrast levels, and data volume reduction in a video stream. It is shown that the perfect organization of the human retina being used as a prototype allows to significantly improve technical characteristics of computer vision systems. The results of this research were used in creating a family of intelligent video cameras and a number of systems based on them and also in constructing specialized neural networks for primary information processing directly on a sensor matrix. Keywords: retina, rod cell, cone cell, horizontal cell, bipolar cell, amacrine cell, ganglion cell, on-center, off-center, neural network, video sensor, control of parameters of reading information, intelligent video camera, multilayer matrix structure. INTRODUCTION Real-time video systems play a significant role in systems of automation of production processes, visual quality control of products, robotics, defensive and military systems, systems of automation of scientific and biomedical research, etc. The range of application of these systems and requirements on them are constantly extended. This is applied in particular to video feedback systems in which the results of real-time information processing are used to control a process or other actions. Such systems must satisfy increased requirements not only on the performance of computing facilities but also on the time delay of information in feedback loops, which are not ensured within traditional approaches. The human visual system has been polished up over millions of years and has reached an extremely high level of organization. A generalized model of the human visual system is multifunctional and is composed of several tens or even a hundred of local models that describe a number of structural, physical, geometric, and psychophysical mechanisms and processes. The process of perception of visual information by a person is dynamic with many changeable parameters and feedbacks. We not only look but also see and react, i.e., this process is active. Therefore, the phenomenon of human vision provides an extremely large number of diverse elegant solutions for implementing computer vision systems. Since the material of this article is at the junction of several specialties, to better understand it, we will give a short description of the organization of the human eye retina as applied only to processes of perceiving images and processing video information on neuron layers [1, 2]. 1
V. M. Glushkov Institute of Cybernetics, National Academy of Sciences of Ukraine, Kyiv, Uk
Data Loading...
