Re-weighting features selection based on wrapper filter and genetic algorithms for figurative images recognition

This paper describes a feature selection technique for logo recognition to deal with a large number of images in the trademark registration system. The proposed approach is based on genetic algorithms with features re-weighting process. Diverse types of features are combined to perform an evolutionary process applied on four logotypes. Comparative results show that our method can support invariance in partial image deformations and outperforms other existing techniques that use one feature alone.