HMM-Based Online Recognition of Handwritten Chemical Symbols

A SVM-HMM Based Online Classifier for Handwritten Chemical Symbols

“…As a stochastic model, HMM are very suitable for handwritten symbols and characters recognition [9]. Based on the first-stage classification result, we mainly adopt the method in our previous work [5] to build hmm classifiers and expand it to the classification of ORS Symbols. To improve the accuracy of ORS symbols, we design a PSR algorithm to preprocess samples before feature extraction.…”

“…We mainly adopt the preprocessing and features used in our previous work [5]. Especially, to improve the accuracy of ORS, we use PSR to preprocess the samples of ORS before feature extraction.…”

“…Xin [4] proposed an approach to understand handwritten chemical formulas. In our preliminary work [1] and [5], we proposed the system architecture for handwritten chemical formulas recognition, and used hmm-based method to recognize inorganic symbols and achieved 89.5% top-1 accuracy.…”

Computer Vision Techniques for Hidden Conditional Random Field-Based Mandarin Phonetic Symbols I Recognition

“…MPS1 symbols have long range dependent features so this work modifies the potential function ) in Eq. 9 to add a window parameter s w [13]. The parameter s w defines the amount of past and future observations.…”

“…This system extracts twelve-dimensional features in each stroke. The features are the stroke length ratio feature, the horizontal stroke feature, the vertical stroke feature, the stroke based loci features, and the stroke curvature feature [13]. These features are computed as follows:…”