Object recognition using characteristic component and genetic algorithms

Object recognition is an essential part of the computer vision system. This paper uses a genetic algorithm to select a model shape that has the best match with invariant input images. The contour shape of an image is described in terms of shape features such as straight lines, curves, and angles. In the first step of the method, the shape feature is identified by analyzing the contour, and measuring the invariant properties of the normalized features. The second step obtains coding of the shape features as attributed strings and stores this in the database of the system. Finally, the procedure in the first and second steps is used to obtain the input model and uses a genetic algorithm to find the best-matched model with an input model by searching the best-matched model from the database. From this method we can recognize an unknown object. The algorithm is tested with 20 objects rotated in different orientations. The results are encouraging, since we achieved 95.9% correct recognition.

[1]  D. E. Goldberg,et al.  Genetic Algorithms in Search , 1989 .

[2]  Pisit Phokharatkul,et al.  Handwritten Thai Character Recognition Using Fourier Descriptors and Genetic Neural Networks , 2002, Comput. Intell..

[3]  Paul M. Griffin,et al.  Point pattern matching using centroid bounding , 1989, IEEE Trans. Syst. Man Cybern..

[4]  King-Sun Fu,et al.  Shape Discrimination Using Fourier Descriptors , 1977, IEEE Trans. Syst. Man Cybern..

[5]  Ender Özcan,et al.  Shape recognition using genetic algorithms , 1996, Proceedings of IEEE International Conference on Evolutionary Computation.

[6]  Luis Enrique Sucar,et al.  Translation, rotation, and scale-invariant object recognition , 2000, IEEE Trans. Syst. Man Cybern. Part C.