A hybrid approach of genetic algorithms and fuzzy logic applied to feature extraction from multisensory images

In this paper, the concept of genetic algorithms (GAs) is introduced to compensate for some undesirable properties that are inherited in the use of fuzzy logic algorithms. The use of GA enables a designer to automatically generate membership functions which are necessary for modeling uncertainty associated with given distributions. This hybrid approach of fuzzy logic and GAs is applied to solve a typical image processing problem. The extracted features from different image modalities (i.e., range and intensity images) are used for image segmentation. This technique yields improved results when compared with those based solely on fuzzy logic systems. Results obtained from this work are examined and evaluated.