A New Type of ART2 Architecture and Application to Color Image Segmentation


Ai, J., Funt, B. and Shi, L., "A New Type of ART2 Architecture and Application to Color Image Segmentation," 18th International Conference on Artificial Neural Networks(ICANN), 89-98, 2008


Abstract:

A new neural network architecture based on adaptive res- onance theory (ART) is proposed and applied to color image segmen- tation. A new mechanism of similarity measurement between patterns has been introduced to make sure that spatial information in feature space, including both magnitude and phase of input vector, has been taken into consideration. By these improvements, the new ART2 archi- tecture is characterized by the advantages: (i) keeping the traits of classi- cal ART2 network such as self-organizing learning, categorizing without need of the number of clusters, etc.; (ii) developing better performance in grouping clustering patterns; (iii) improving pattern-shifting problem of classical ART2. The new architecture is believed to achieve e?ective unsupervised segmentation of color image and it has been experimentally found to perform well in a modi?ed L*u*v* color space in which the per- ceptual color di?erence can be measured properly by spatial information.


Full text (pdf)


Keywords: ART2, similarity measurement, unsupervised segmentation, color image.


Back to SFU Computational Vision Lab publications (home)