Image thresholding (as the simplest form of segmentation) is a very challenging task because of the differences in the characteristics of different images such that different thresholds may be tried to obtain maximum segmentation accuracy. In this paper, a supervised neural network is used to "dynamically" threshold images by assigning a suitable threshold to each image. The network is trained using a set of simple features extracted from medical images randomly selected form a sample set and then tested using the remaining medical images. The results are compared with the Otsu algorithm and the active shape models (ASM) approach. © 2010 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Othman, A. A., & Tizhoosh, H. R. (2010). A neural approach to image thresholding. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6352 LNCS, pp. 561–564). https://doi.org/10.1007/978-3-642-15819-3_72
Mendeley helps you to discover research relevant for your work.