Discussion About Object Profile Cognitive Based On Biological Characteristics

Object profile cognitive is an important area of cognitive psychology and artificial intelligence, it is a subject combine such as physiological vision, cognitive science, psychology, computer science and etc。The robot image understanding technology based on biological characteristics is the forefront technology of the medium -and long-term of our national science and technology development plan。This paper studies the biological visual perception system, discussed the structure of the retina and working mechanism, Lateral inhibition mechanism and the main visual cortex neurons and their receptive field。Then explore some of the research achievements of visual cognitive psychology, such as the law of visual homeomorphic and the graphics structural advantages effect, and discussed the physiological basis of graphic structural advantages effects。The design of this robot visual perception system is based on this theoretical basis。This paper describes the design of the robot vision system at five-layers。First is the photo receptor cell layer, and the photo receptor cell has the perceive function of image brightness information。The second is the profile detection layer, designed mainly the bipolar cells, horizontal cells and ganglion cells, and the receptive field of these neurons, this layer realized the perceptive of object profile。The third is the preliminary cognitive layer, designed super-cells to calculate the topological properties of object profile, then, use this basis for classification。The fourth is the character detect layer, it design simple cells and super-complex cells to detect the Corner of a graphic profile。Fifth layer is the judgment and cognitive layer, it integrated the topological properties and the corner information, then, this system compared with the computer set of experiences and realize the cognitive of external input。This paper also discusses the mathematical structure of neurons in this cognitive system, these mathematical structures can classify the simulated neurons and classify these neurons and their links, these can better describe visual cognitive system and the transfer of information。The main conclusions are:1, The photo receptor's photochemical reaction incidents constitutes incidents aggregate and so is pixel brightness value changes incidents, and those two aggregates have the structure of circulation group。2, Photo receptor cells in software In Fovea have the structure of discrete topological space。And with the relative of location between cells it has the structure of category。3, Bipolar cells and their receptive fields in software In Fovea have the structure of discrete topological space, And this two topological space homeomorphous。4 Quotient aggregate Ganglion /R1 constitutes discrete topological space, and this topological space is the Quotient topological space of ( Ganglion,T G)。5, Ganglion cells in software In Fovea constitutes topological space T G and super cells can constitutes topological space T Super。These two topological spaces are homeomorphous。Finally, do a large number of experiments to test this system five layers function。Show the object profile cognitive based on the biological characteristics is effective。The academic value and practical significance of this paper: this paper integrated cognitive science, psychology, computer science, mathematic and other disciplines of modern achievements, and put forward a set of method to object profile cognitive based on biological characteristics。This method avoid use algorithms, it use some effective mechanism in biological systems for problem's solution and get a well experiment result。This paper shows that research based on biological characteristics of the profile cognitive theory and technology is indeed effective, it is worth further studies。...