An Image Segmentation Algorithm Based on Formal Concept Analysis
Fang Dai, Jiaqi Niu · 2023
Image segmentation is a crucial step from image processing to image analysis, which is extensively applied in image understanding and pattern recognition. Formal concept analysis describes the relationship between objects and attributes by defining formal concepts, which is a classic mathematical analysis method. In recent years, formal concept analysis has been applied to analyze social networks, but its application in image networks is relatively vacant. In this paper, we propose an image segmentation algorithm based on formal concept analysis. First, the image is transformed into a network structure using our proposed network construction method based on the local density of nodes. Then, the image network is clustered using the k-clique community detection algorithm based on formal concept analysis. The nodes in a community represent the homogeneous regions of the image. Thus, the results of image segmentation are achieved by merging these homogeneous regions. We conducted experiments on the Berkeley Segmentation Dataset BSDS500. The experimental results show that the proposed algorithm of image segmentation in this paper performs better than several existing algorithms.