CGFFCM: A color image segmentation method based on cluster-weight and feature-weight learning
Amin Golzari Oskouei, Mahdi Hashemzadeh
Abstract
CGFFCM (Cluster-weight and Group-local Feature-weight learning in Fuzzy C-Means) is a clustering-based color image segmentation approach. It applies an automatic cluster weighting strategy to mitigate the initialization sensitivity and a group-local feature weighting technique to improve the clustering accuracy. In addition, it exploits an efficient combination of image features, consisting of eight features from three different groups (i.e., local homogeneity, CIELAB color space, and texture), to increase the image segmentation quality. CGFFCM also utilizes the imperialist competitive algorithm to optimize its feature weighting process. An open-source Matlab implementation of CGFFCM is available.
Topics & Concepts
Artificial intelligencePattern recognition (psychology)WeightingCluster analysisComputer scienceImage segmentationFeature (linguistics)InitializationComputer visionImage textureSegmentationLinguisticsRadiologyPhilosophyProgramming languageMedicineRemote-Sensing Image ClassificationRemote Sensing and Land UseImage Retrieval and Classification Techniques