next up previous
Next: Limitations of thresholding Up: Region-based segmentation Previous: Region-based segmentation


Thresholding is the simplest way to perform segmentation, and it is used extensively in many image processing applications. Thresholding is based on the notion that regions corresponding to different tissue types can be classified by using a range function applied to the intensity values of image pixels. The assumption is that different tissue types will have a distinct frequency distribution and can be discriminated on the basis of the mean and standard deviation of each distribution (see Figure gif). For example, given a two-dimensional image I(x,y) , we can define a simple threshold rule to classify bony tissue or a compound threshold rule to classify fatty tissue:

Figure: A hypothetical frequency distribution f(I) of intensity values I(x,y) for fat, muscle and bone, in a CT image. Low intensity values correspond to fat tissues, whereas high intensity values correspond to bone. Intermediate intensity values correspond to muscle tissue. F+ and F- refer to the false positives and false negatives; T+ and T- refer to the true positives and true negatives.

Thresholding has been used by Chow and Kaneko to segment ventricles from cineangiograms of the human heart [18]. They chose this technique because of the strong bimodal distribution of image values corresponding to regions that are interior and exterior to the ventricles.

Ramani Pichumani
Mon Jul 7 10:34:23 PDT 1997