Feature extraction
From Wikipedia, the free encyclopedia
In pattern recognition and in image processing, Feature extraction is a special form of dimensionality reduction.
Contents |
[edit] General
In many problems the number of variables is very large. This can mean that processing of the data is slow, requires a lot of memory or that classification algorithm overfits to the training examples, thus generalizing poorly to new samples. Feature extraction is a general term for methods for constructing combinations of the variables which get around above problems but still describe the data sufficiently accurately.
Best results are achieved when an expert constructs a set of application-dependent features. Nevertheless, if no such expert knowledge is available general dimensionality reduction techniques may help. These include:
- Principal components analysis
- Semidefinite embedding
- Multifactor dimensionality reduction
- Nonlinear dimensionality reduction
- Isomap
- Kernel PCA
- Latent semantic analysis
- Partial least squares
[edit] Image processing
It can be used in the area of image processing which involves using algorithms to detect and isolate various desired portions or shapes (features) of a digitized image or video stream. It is particularly important in the area of Optical Character Recognition.
[edit] Low-level
[edit] Curvature
- Edge direction, changing intensity, autocorrelation.
[edit] Image motion
- Motion detection. Area based, differential approach. Optical flow.
[edit] Shape Based
[edit] Thresholding
[edit] Template matching
[edit] Hough transform
- Lines
- Circles/Ellipse
- Arbitrary shapes (Generalized Hough Transform)
[edit] Flexible methods
- Deformable, parameterized shapes
- Active contours (snakes)
