期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：24
页码：3692-3697
出版社：Journal of Theoretical and Applied
摘要：Due to the great influence in image analysis and processing systems, the problem of image segmentation has been studied for years; there is not yet any automatic method able to correctly process any kind of image. In this study we did concentrate about the most important method of segmentation, �The Watershed transformation�. However, The watershed method has interesting properties that make it useful for many different image segmentation applications: it is simple and intuitive can be parallelized, and always produces a complete division of the image, when applied to medical image analysis, it has important drawbacks: over segmentation, sensitivity to noise, poor detection of thin or low signal to noise ratio structure and others problems, the current approach we use due to correct some drawbacks, to minimized the number of watersheds by merging it but is still close to the original.
关键词：Image; Watershed Transform; Segmentation; Merging Region; Cost; Variance; Nodes; Region
