Arakawa Kaoru    Department   Undergraduate School  , School of Interdisciplinary Mathematical Sciences    Position   Professor
Language	English
Publication Date	2012/07
Type	International Conference
Title	Complex Rule-Based Filter for Impulsive Noise Reduction in Color Images Optimized by Interactive Evolutionary Computing
Contribution Type	Co-authored (other than first author)
Journal	Proc. ITC-CSCC 2012
Volume, Issue, Page	pp.D-T2-04
Author and coauthor	Yohei Katsuyama and Kaoru Arakawa
Details	A method for removing impulsive noise from color images which utilizes noise detection with interactive evolutionary computing (IEC) is proposed. This system is realized as a complex rule-based filter composed of median filters and an interpolative technique. The Median filter is good for removing impulsive noise but tends to make the image blurred, while the interpolative filter can avoid blurring, but cannot remove high-density noise well. Effective noise removal can be realized by switching them depending on how the noise is added. Since the performance is influenced by the accuracy of noise detection, multiple rules with parameters are applied in the system. In order to optimize these parameters, IEC is adopted. IEC can also consider human subjective criteria to image quality. Computer simulations show the high performance of this system in actual noise reduction for color images.