This paper proposes the framework of an image browsing method, which puts its importance on an exploratory search based on the interaction between users and computers. In comparison to the previous works, the characteristics of our method is the application of evolutionary computation to the interactive optimization of similarity measures. The fitness of each individual is related to the value of importance feedback given by a user to images retrieved in higher orders. The result of image browsing on satellite cloud images show that keeping population-based multiple similarity measures in evolutionary computation works effective for image browsing.

      @InProceedings{ kt:iim98,
      author = {Asanobu KITAMOTO and Mikio TAKAGI},
      title = {Interactive Image Browsing Method Based on Evolutionary Computation},
      booktitle = {Proceedings of the 4th Symposium on Intelligent Information Media},
      pages = {173-180},
      year = 1998,
      month = 12,
      doi = {},
      note = { (in Japanese)},
      }