Title :
Extraction of disconnected closed curve using genetic algorithm with perceptive grouping factors
Author :
Saitoh, Fumihiko
Author_Institution :
Cores Corp. Co. Ltd., Tokyo, Japan
fDate :
29 Sep-1 Oct 1997
Abstract :
A method for extracting a visual figure of a disconnected and arbitrary closed curve in an image is proposed using a genetic algorithm based on factors of a closeness and a continuity in a perceptive grouping functions. An array of bits which represents all pixel numbers in an image is defined as the genotype of an individual. The length of a gene information is fixed and the number of pixels to compose a closed curve is not limited. Therefore a proper closed curve can be extracted even if it has complicated form and a long contour. As the experiment result, visual proper closed curves were extracted in noisy images efficiently by the method
Keywords :
computer vision; edge detection; feature extraction; genetic algorithms; computer vision; disconnected closed curve; feature extraction; gene information; genetic algorithm; perceptive grouping factors; visual figure; Data mining; Genetic algorithms; Genetic mutations; Humans; Image recognition; Joining processes; Machine vision; Pixel; Psychology; Scattering;
Conference_Titel :
Robot and Human Communication, 1997. RO-MAN '97. Proceedings., 6th IEEE International Workshop on
Conference_Location :
Sendai
Print_ISBN :
0-7803-4076-0
DOI :
10.1109/ROMAN.1997.646991
