Morph Segmentation Flashcards
How do you define Morph Segmentation?
- operates on images as a set of points
- modify in a control way the morphology of an image
- used in binary images
- used in image processing for filtering, segmentation, object description
It has basic and composed operations
How many basic operations are there?
2
What are the basic operations that morph segmentation allows?
- dilation
- erosion
How do you define dilation?
- enlarges the boundaries
- small holes and gaps are filled
How do define erosion?
- erode away the boundaries of regions of foreground
- holes and gaps are increased
How many composed operations are there?
2
What are the composed operations that morph segmentation allows?
- opening
- closing
What characterizes opening?
- erosion followed by dilation
- union of all objects that fit in structuring element
What characterizes closing?
- dilation followed by erosion
- dual to opening
Can edge detection be done with morphology? If yes how does it work?
Yes
- erosion of objects then subtraction from original
How is segmentation done?
- dividing image in regions
- applied before image analysis
- group pixels with similar properties
- no perfect segmentation method
What is edge detection segmentation based in?
- pixel intensity
- regions
- edges
What is thresholding?
- oldest segmentation method
- appropriate when object of interest have homogenous intensity different from background
- not easy to find the adequate value
- if value is uknown, analyze the histogram to choose an adequate value
-For a bimodal histogram, the threshold value corresponds to the valley between the peaks
Can this approach can produce classification errors depending on the image histogram and intensity of values of the objects?
Yes
How many and what kind of approaches are there (thresholding)?
-There are 2 approaches:
- global
- variable
How many variable approaches are there?
3
-Local
-Adaptive
- Otsu’s method
Give me a example of thresholding
OCR -Optical Character Recognition:
Page ->Scanner -> OCR -> ASCII text
Can the thresholding be applied to image
sub-regions?
Yes
What defines the local variable approach?
It depends on properties of neighbouring pixels
What defines the Adaptive variable approach?
It depends on spatial coordinates
What defines the Otsu’s method variable approach?
It is based on probabilistic analysis obtained
from histogram
What is region growing:
- grow a region by aggregation of pixels starting at a seed point
- all neighbouring pixels that comply the rule are labelled as belonging to the region
Is obtaining a “good” seed pixels a problem ?
Yes
How can seed be obtained?
Seed can be obtained using:
– Histograms
– Interactively
…
How can flood filing be done?
It can be done using:
-recursive version;
-depth-first search;
-breadth-first search;