Non-linear regression based feature extraction for connected-word recognition in noise.
Frank SeideAlfred MertinsPublished in: ICASSP (2) (1994)
Keyphrases
- word recognition
- feature extraction
- noisy environments
- cursive script
- handwritten word recognition
- character recognition
- handwriting recognition
- word segmentation
- feature vectors
- document analysis
- information retrieval
- speech recognition
- image processing
- noise reduction
- pattern recognition
- historical documents
- signal to noise ratio
- automatic speech recognition
- feature space
- connected components
- feature selection
- video analysis
- dimensionality reduction
- wavelet transform
- machine learning