Paper Title
Intelligent Control Based Multilevel And Multiband Digital Image Thresholding Using Fuzzy Entropy And Particle Swarm Optimization
Abstract
Thresholding is a fundamental image segmentation approach which transforms grayscale images into binary
images. For more complicated cases than bi-level thresholding, extension is necessary in order to fulfill multilevel and
multiband image thresholding. In multilevel thresholding, computation complexity increases significantly. In multilband
thresholding, multiple dimension optimization should be made across different color frequency bands. Balanced histogram
thresholding is widely accomplished in automatic image thresholding. Fuzzy entropy is thus introduced as the objective
function to solve the optimal thresholds, where the image histogram is partitioned into various objects via fuzzy partition, so
that diverse fuzzy membership functions can be assigned. Fuzzy entropy based optimization has been formulated
accordingly. To speed up the convergence rate and shorten the computation time, particle swarm optimization based
intelligent control schemes have been proposed to optimize thresholds rapidly. Comparisons of the role of fuzzy membership
assignment on the thresholding quality are also made. Superior outcomes on multilevel and multiband thresholding are
reached.
Index Terms— Fuzzy Entropy, Fuzzy Membership Function, Particle Swarm Optimization, Intelligent Control, Multilevel
Thresholding, Multiband Thresholding