Tuesday, February 19, 2008

Letter Of Not Renewing Labour Contract

image segmentation

for the hand of all the other things, such as filter, for example, background, etc. can be taken to segment the hand first.




There are various methods for image segmentation that can be applied.
The boundaries between these processes is often blurred. Therefore, one can combine them to achieve better and efficient results.

Pixel-oriented method

The pixel-oriented methods, criteria used to segment allocation only to the pixels, independently of other pixels.

Procedure: Schwelwertverfahren

edge oriented method

The edge-oriented method is searched for edges and contours, which are combined to object boundaries.

methods: Sobel operator, Laplace operators, watershed transformation, Snakes

region-oriented method

The region-based methods are considered contiguous sets of points as a whole.

methods: region growing, split and merge, Pyramid Linking, Region Merging

model-based method is

The model-based methods based on concrete knowledge about the pictures, that is a model.

Procedure: Hough transform, statistical models, template matching

texture-based method

in texture-based method for a texture, ie, after a homogeneous internal structure, rather than segmented according to a uniform color.

Procedure: cooccurrence matrix (Haralick matrices),
texture energy measures (Texture Energy Measure),
Lauflängenmatrizen (Run-Length Matrix),
fractal dimensions and measures,
Markov Random Fields and Gibbs potentials,
structural approaches, theoretical concepts
signal

These methods are sometimes on the borderline for classification or allow simultaneous segmentation and classification.


0 comments:

Post a Comment