Improving Handwritten OCR with Training Samples Generated by Glyph Conditional Denoising Diffusion Probabilistic Model.
Haisong DingBozhi LuanDongnan GuiKai ChenQiang HuoPublished in: ICDAR (4) (2023)
Keyphrases
- training samples
- denoising
- probabilistic model
- character recognition
- feature space
- image denoising
- nonlinear diffusion
- optical character recognition
- high dimensional
- test sample
- training data
- supervised learning
- training set
- number of training samples
- learning algorithm
- hyperplane
- character segmentation
- image processing
- bayesian networks
- handwriting recognition
- face images
- decision function
- post processing
- sample set
- discriminative information
- machine vision
- document analysis
- decision boundary
- natural images
- text lines
- weak classifiers
- anisotropic diffusion
- data sets
- document images
- representative samples