DocumentCode
3235940
Title
Evolutionary mechanism and implemention for recognition of objects in dynamic vision
Author
Sha, Sha ; Jianer, Chen ; Ling, Qin ; Sanding, Luo
Author_Institution
Sch. of Inf. Sci. & Eng., Central South Univ., Changsha, China
fYear
2009
fDate
25-28 July 2009
Firstpage
178
Lastpage
182
Abstract
To improve the real-time and robustness of the dynamic visual recognition, this paper proposed a new mechanism and implementation techniques for evolutionary recognition, which is a perceptive--expressive interaction imitating the oil painting process. This mechanism can transform the oil painting process of continual observation, multi-level description, and repeated confirmation into the gradual maturity process of evolutionary objects driven by the force of evolution, as well as analyze the evolution process of evolutionary objects. Also, this paper put forward an evolution model of the evolutionary object as well as three kinds of evolution driving force, discussed the key technologies of evolutionary recognition, verified the feasibility, as well as raised questions that need to be resolved through the experiment of road environmental objective recognition.
Keywords
computer vision; evolutionary computation; object recognition; dynamic vision; dynamic visual recognition; evolutionary mechanism; gradual maturity process; object recognition; oil painting process; perceptive-expressive interaction; road environmental objective recognition; Biological system modeling; Evolution (biology); Evolutionary computation; Genetics; Image recognition; Object oriented modeling; Object recognition; Painting; Petroleum; Roads; computer vision; evolution machenisim; evolutionary vision process; multi-level classes; object recognition;
fLanguage
English
Publisher
ieee
Conference_Titel
Computer Science & Education, 2009. ICCSE '09. 4th International Conference on
Conference_Location
Nanning
Print_ISBN
978-1-4244-3520-3
Electronic_ISBN
978-1-4244-3521-0
Type
conf
DOI
10.1109/ICCSE.2009.5228497
Filename
5228497
Link To Document