General image segmentation is a non-trivial task, which requires significant computational power and huge amount of knowledge incorporated. Fortunately, it is not necessary in all the cases. In some specific cases, simpler non-supervised or supervised segmentation methods can be used giving even better results. In this paper, a novel trainable segmentation method based on RapidMiner data-mining platform is introduced, and its functionality is described. The method implementation was released under open-source license as a part of IMMI (IMage MIning) extension of the RapidMiner platform. When compared to other trainable segmentation algorithms, the platform provides flexibility connected with all the features of one of the most widely used data-mining platform today. The functionality has been verified on the satellite image use-case, accuracy achieving 78.3% pixel error. © Springer-Verlag Berlin Heidelberg 2013.
CITATION STYLE
Masek, J., Burget, R., & Uher, V. (2013). IMMI: Interactive Segmentation Toolkit. In Communications in Computer and Information Science (Vol. 383 CCIS, pp. 380–387). Springer Verlag. https://doi.org/10.1007/978-3-642-41013-0_39
Mendeley helps you to discover research relevant for your work.