Model-based recognition in robot vision
ACM Computing Surveys (CSUR)
Genetic Algorithms in Search, Optimization and Machine Learning
Genetic Algorithms in Search, Optimization and Machine Learning
Computer Vision
Hi-index | 0.00 |
Object recognition is an essential part of the computer vision system. This paper proposes a genetic algorithm to search the features of model shapes of the object from model-base, to identify input shapes of the object. The dominant points are extracted from the edge of binary images using Gaussian filtering. There are two methods to compute the output features. The first, B-Spline, used the dominants to compute the control points. The second, Cardinal Spline computes the data points form the dominant points. The control points, and the data points are built a model shapes for searching by genetic algorithms to identify the input images. Then, we are compared the two method. Training data composes of original object, its translation, its rotation and its scaling. The recognition results of B-Spline implementation are 97% for rotated object, 94% for rotated and scaling object. The recognition results with Cardinal Spline feature are 97% for rotated, 95.2% for rotated and scaling object.