Hierarchical segmentation using α -trees can suffer from unwanted leakage or chaining effects, which lower segmentation quality by reducing the depth of the hierarchy. In this paper we introduce a new way to prevent the chaining effect of α -trees. It relies on the odd 2-D Gabor filter. A series of clean, noisy, and blurred synthetic images was used to test the ability of improved α -tree to stop the chaining effect obtaining a 99.8% segmentation accuracy, and we compared it with the contrast-based α -tree proposed previously by Soille. Two remote sensing images were also used to test the performance of the methods on natural images. The results showed that both 4-CN and 8-CN odd Gabor filter based α -trees can prevent the chaining effect efficiently.
CITATION STYLE
Zhang, X., & Wilkinson, M. H. F. (2019). Preventing chaining in alpha-trees using gabor filters. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11564 LNCS, pp. 268–280). Springer Verlag. https://doi.org/10.1007/978-3-030-20867-7_21
Mendeley helps you to discover research relevant for your work.