Research On Models Of Visual Information Processing Based On Central Neural Mechanisms Of Vision

Visual information processing model is a comprehensive and challenging subjeet, it includes computer science, nerve science and perceive science. So far, plentiful fruits have been got in that research. However,people also find that the difficulty is much greater than what's expected in the process of exploring visual perception more. For crossing the road as an example, thouge now the ability of computer is so supernatural, computer visual can not guide this task. However the child before school age can complete this task easily. This shows that the difference between computer vision and human vision. That difficulty roots in the localization of the understanding of visual mechanism and technical restrictions to visual computing. So, people must improve their understanding of the vision and renovate their computing means in order to solve the problems. Study results indicate that the human visual information processing is a highly intelligent and highly adaptive, the main reason lies in the level of processing and parallel. Therefore, the simulation of the characteristics of the brain, constructed parallel information processing level model, visual information processing model is the core issue.Progress in the study of the visual neural make people more in-depth understanding of the principles of biological vision, but also for computer vision research has brought new vitality. Human visual multi-channel, parallel processing model will have a more in-depth study, including color information processing channel, color characteristics of independence, as well as the color of geometry features, the translation, rotation, scaling transformation is not deformation , the size of the image itself, direction, the dependence on the angle, showing the robustness of a strong advantages, and receive much attention. By using color image retrieval features of the three key issues: the color space, color feature extraction and color-based similarity measure. Which color space used in this paper that with the characteristics of the human eye HSV space, and its non-uniform quantization, quantification of 24 colors, that is, the retention of the original image, and greatly improving the calculation speed; Color feature extraction based on the use of space and color histogram autocorrelation ways to avoid the traditional histogram neglected space and location information for regional domination color histogram problem.Content-based image retrieval is computer vision, pattern recognition, database management systems, artificial intelligence and other subjects a product of integration. In today's digital society, the rapid growth of image data, far more than the existing system of retrieval capabilities, data from a large number of images quickly satisfy user needs to retrieve images is of great practical significance. This paper describes a color feature extraction model at the same time as the basis for design of the space and color histogram-based image retrieval systems, experimental results show that the retrieval system is a high efficiency.Based on the central nervous system as the basis of color information to focus on the visual information processing model for in-depth studies, the specific work of this paper include the following:â‘ Summarize the visual information processing model of the history, status quo, and the problems faced by the visual neural mechanisms principles, methods and simulation strategy and other basic knowledge, these visual information processing model is the foundation.â‘¡Based on the central nervous system, visual information processing model of the human visual model was proposed. experiments was designed to simulate retinal the process of information processing, generate the corresponding retinal images, the feelings of simple cell retinal field again for the input image, through stimulation of specific direction, and caused many edge information, and formed a contour edge image. The experimental results show that: the model have a good simulation.â‘¢A space and color Co-occurrence histogram related feature extraction model was proposed, the model used HSV color space which in accord with characteristics of the the human eye, using its non-uniform quantify technology to reduce the colour dimension, improved retrieval efficiency. Since only considering the same color among the image, reducing the computational complexity of the time. Space-based model is more concerned about the color space layout, in the histogram can be embodied in the spatial position information, and more emphasis on the importance of regional color transition, in order to avoid regional dominance in a row color histogram problem. Histogram intersection algorithm used to calculate the image similarity to avoid the standard L1 distance demands an exact bin-by-bin match between histograms.â‘£Based on the color characteristics, a image retrieval system was designed, which is in the Windows XP operating system, the development of tools using Matlab, uses containing 11 categories of 220 images as detection database, the system used in Chapter 4 of the space and color Co-occurrence histogram method to extract the color feature. The experimental results show that: the system has a higher query accuracy rate and query Integrity rate.Given the level of my ability, there are still many places to improve, I have the following prospects for the next step in the field:â‘ This image retrieval system based on the characteristics of color, if we can add shape, texture and other information to the system, the search results will be increased.â‘¡Primary visual cortex cells extract the basic characteristics of images, these characteristics then arrived at the high visual cortex. How to integrate and how to store these features, as well as the attention mechanism and association mechanism of high visual cortex are worthy of in-depth research on the subject.â‘¢All the visual features in the primary visual cortex are integrated, stored, when the stimulation emergences of the outside world, people will response,this is human's attention mechanism and association mechanism. In thinking computation, image direct knowledge make tremendous contributions to image thingking. How to construct mental imagery and how to construct the network based on mental imagery are the problem in the next step .The future of visual information processing model is optimistic, with the development of cognitive psychology and neural science, the gap between computer vision and human vision is shortened, I believe that in the near future, computer vision will be able to competent human vision.