{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

# ch10b - Ch10 Segmentation Overview Thresholding Region...

This preview shows pages 1–8. Sign up to view the full content.

Ch10 – Segmentation • Overview • Thresholding Region Growing Split and Merge Watershed Segmentation • Conclusion

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Overview The purpose of segmentation is to divide an image into “visually sensible” regions corresponding to objects of interest Segmentation is the “dual” of edge detection where we focus on finding object boundaries In most cases, the pixels inside a region are homogeneous (same brightness or color)
Overview Different segmentation algorithms have been devised to group similar pixels together to form regions for each object of interest – Global – thresholding – Bottom up – region growing – Top down – split and merge – Geometric – watershed Once regions are found, we often need to calculate their properties (size, position, shape)

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Thresholding The basic idea of image thresholding is to divide the intensity range in an image into two categories based on a threshold value T – Background = {0..T-1} – Object = {T..255} The key is finding the threshold value T that produces the most sensible results
Thresholding Many interactive image processing systems allow users to select thresholds manually Some systems use a priori domain knowledge to select appropriate thresholds (eg. fax) Most automatic threshold selection methods are based on the intensity distribution of the input image and optimize some quality metric

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Thresholding • The mean midpoint threshold algorithm selects a threshold value T to be the midpoint between the mean object intensity and the mean background intensity – Guess initial T – Calculate M 1 = mean < T – Calculate M 2 = mean > T – Update T=(M 1 +M 2 )/2 – Iterate until convergence
Thresholding original image threshold image histogram

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}