Improving Off-line Handwritten Character Recognition with ...

Improving Off-line Handwritten Character Recognition with ...

Handwritten Character Recognition using Hidden Markov Models Quantifying the marginal benefit of exploiting correlations between adjacent characters and words Optical Character Recognition Rich field of research with many applicable domains Off-line vs. On-line (includes time-sequence info) Handwritten vs. Typed

Cursive vs. Hand-printed Cooperative vs. Random Writers Language-specific differences of grammar and dictionary size We focus on off-line mixed-modal English data set with mostly handwritten and some cursive data Observation is monochrome bitmap representation of each letter with segmentation problem already solved for us (but poorly) Pre-processing of dataset for noise filtering and normalizations of scale also assumed done Common Approaches to OCR Statistical Grammar Rules and Dictionaries Feature Extraction of observations Global features: Moments and invariants of image (e.g., percentage of pixels in certain region, measuring curvature) Local features: Group windows around image pixels Hidden Markov Models Used mostly in cursive domain for easy training and to avoid segmentation issues

Most HMMs use very large models with words as states, combined with above approaches, which is more applicable to domains of small dictionary size with other restrictions Visualizing the Dataset Data Collected from 159 subjects with varying styles, printed and cursive Missing first letter of each word to simplify capital letters Each character represented by 16x8 array of bits Character meta-data includes correct labels and

end-of-word boundaries Pre-processed into 10 cross-validation folds Our Approach: HMMs Primary Goal: Quantify the impact of correlations between adjacent letters and words Secondary Goal: Learn an accurate classifier for our data set Our Approach: Use a HMM and compare to other algorithms 26 states of HMM each represent letter of alphabet Supervised learning of model with labeled data Prior probabilities and transition matrix learned by frequency of letters in training Learning algorithm for emission probabilities uses Naive Bayes assumption (i.e., pixels conditionally independent given the letter) Viterbi algorithm predicts most probable sequence of states given the observed character pixel maps Algorithms and Optimizations

Learning algorithms implemented and tested: Baseline Algorithm: Nave Bayes Classifier (no HMM) Algorithm 2: NB with maximum probable classification over a set of shifted observations Algorithm 3: HMM with NB assumption Fix for incomplete data: Examples hallucinated prior to training Algorithm 4: Optimized HMM with NB assumption Motivation was to compensate for correlations between adjacent pixels not included in Nave Bayes assumption

Ignore effects of inter-word transitions when learning HMM Algorithm 5: Dictionary Creation and Lookup with NB assumption (no HMM) Geared toward specific data set with small dictionary size, but less generalizable to more constrained data sets with larger dictionaries Alternative Algorithms and Experimental Setup Other variants considered but not implemented: Joint Bayes parameter estimation (too many probabilities to learn, 2^128 vs. 3,328)

HMM with 2nd-order Markov assumption (exponential in number of Viterbi paths) Training Nave Bayes over a set of shifted and overlayed observations (preprocessing to create thicker boundary) All experiments run with 10-fold cross-validation Results given as averages with standard deviations Experimental Results 100% 95% 90% Test Accuracy 85% 80% 75% 70% 65% 60% 55% 50% Nave Bayes

NB w/ Shifted Observations Standard HMM Classification Algorithms Optimized HMM Dictionary Creation Conclusions Nave Bayes classifier did pretty good on its own (62.7% accuracy - 15x better than random classifier!) Classification on shifted data did worse since we lost data on edges! Small dictionary size of dataset affected results: Optimized HMM w/ NB achieves 71% accuracy

Optimizations only marginally significant because of dataset More simple and flexible approach for achieving impressive results on other datasets Dictionary approach is almost perfect with 99.3% accuracy! Demonstrates additional benefit of exploiting domain constraints, grammatical or syntactic rules Not always feasible: dictionary may be unknown, too large, or the data may not be predictable

Recently Viewed Presentations

  • Designing Your Poster - Grand Valley State University

    Designing Your Poster - Grand Valley State University

    Designing Your Poster. Making decisions :: Making meaning. ... Gradation of color from warm to cool and of tone from dark to light produce aerial perspective (*showing distance or depth with color or tone -- something farther away will look...
  • Sedimentation and Macro-Invertebrate Species Identification

    Sedimentation and Macro-Invertebrate Species Identification

    An abundance of snail shells, clam shells, caddis fly casings Between shoreline and middle of lake: Few snail shells, few clam shells, chironomid larvae Middle of lake: chironomid larvae, dragonfly larvae, tubifex worms. Results
  • 4He Solubility in Apatite is Low But Possibly Significant

    4He Solubility in Apatite is Low But Possibly Significant

    As part of our efforts to study radiation-damage trapping of 4He in fission tracks, we have been attempting to saturate apatite crystals with helium, and these experiments have given us insight into helium solubility in apatite. Our newest experiments at...

    Totally Integrated Automation 20% besparing op de bekabelingskosten, b.v. op ... Schakelkasten/panelen Installatie tijd De aansluiting Elektrische PROFIBUS FastConnect Systeem: Kabels worden snel en eenvoudig gemonteerd Profibus feiten Snelle en reproduceerbare responstijden ...
  • Following the Steps The Writing Process

    Following the Steps The Writing Process

    For more information on sentence structure and punctuation, see "Sentence Clarity and Combining" and "Conquering the Comma," included on this CD-ROM. For presentations on documentation styles, see "Cross-referencing: Using MLA Format" and "Documenting Sources: Using MLA Format," also on this...
  • Nylon inserts 101 - PART 2 - Continental-Aero

    Nylon inserts 101 - PART 2 - Continental-Aero

    Undersized nylon insert. Physical Appearance: nylon mushrooms or expands upward as the mating end protrudes through the nylon. Possibility of Root Cause: Approximately 8% of nylon issues are considered undersized nylon inner diameter, whereas the opening of the nylon insert...
  • WELCOM Community Meeting E To The SR 134/GLENDALE

    WELCOM Community Meeting E To The SR 134/GLENDALE

    Grind 2" of existing AC. Place 0.5" leveling course. Place . 1.5" ARHM. Begins . NEW. service life . Limits Full AC . Pavement R&R. Parking . and access to your street . may . be unavailable during the day...


    Zoe Lard - Becario. Larry Adkisson - INV. Delitos Ambientales. Michael Melito - SAAG. Troy Arnold - INV. ... Robert Shapiro. Primer Asistente del Procurador . Julie Selsberg. Director. Pete Dunbar. Primer Asistente del Procurador. Sean Clifford.