Neural network based off-line handwritten text recognition system

This dissertation introduces a new system for handwritten text recognition based on an improved neural network design. Most of the existing neural networks treat mean square error function as the standard error function. The system as proposed in this dissertation utilizes the mean quartic error function, where the third and fourth derivatives are non-zero. Consequently, many improvements on the training methods were achieved. The training results are carefully assessed before and after the update. To evaluate the performance of a training system, there are three essential factors to be considered, and they are from high to low importance priority: (1) error rate on testing set, (2) processing time needed to recognize a segmented character and (3) the total training time and subsequently the total testing time. It is observed that bounded training methods accelerate the training process, while semi-third order training methods, next-minimal training methods, and preprocessing operations reduce the error rate on the testing set. Empirical observations suggest that two combinations of training methods are needed for different case character recognition. Since character segmentation is required for word and sentence recognition, this dissertation provides also an effective rule-based segmentation method, which is different from the conventional adaptive segmentation methods. Dictionary-based correction is utilized to correct mistakes resulting from the recognition and segmentation phases. The integration of the segmentation methods with the handwritten character recognition algorithm yielded an accuracy of 92% for lower case characters and 97% for upper case characters. In the testing phase, the database consists of 20,000 handwritten characters, with 10,000 for each case. The testing phase on the recognition 10,000 handwritten characters required 8.5 seconds in processing time.

Reliability of the Clinical Application of a Mechanical Inclinometer in Measuring Glenohumeral Motion

Objective: Establish intra- and inter-examiner reliability of glenohumeral range of motion (ROM) measures taken by a single-clinician using a mechanical inclinometer. Design: A single-session, repeated-measure, randomized, counterbalanced design. Setting: Athletic Training laboratory. Participants: Ten college-aged volunteers (9 right-hand dominant; 4 males, 6 females; age=23.2±2.4y, mass=73±16kg, height=170±8cm) without shoulder or neck injuries within one year. Interventions: Two Certified Athletic Trainers separately assessed passive glenohumeral (GH) internal (IR) and external (ER) rotation bilaterally. Each clinician secured the inclinometer to each subject’s distal forearm using elastic straps. Clinicians followed standard procedures for assessing ROM, with the participants supine on a standard treatment table with 90° of elbow flexion. A second investigator recorded the angle. Clinicians measured all shoulders once to assess inter-clinician reliability and eight shoulders twice to assess intra-clinician reliability. We used SPSS 14.0 (SPSS Inc., Chicago, IL) to calculate standard error of measure (SEM) and Intraclass Correlation Coefficients (ICC) to evaluate intra- and inter-clinician reliability. Main Outcome Measures: Dependent variables were degrees of IR, ER, glenohumeral internal rotation deficit (GIRD) and total arc of rotation. We calculated GIRD as the bilateral difference in IR (nondominant–dominant) and total arc for each shoulder (IR+ER). Results: Intra-clinician reliability for each examiner was excellent (ICC[1,1] range=0.90-0.96; SEM=2.2°-2.5°) for all measures. Examiners displayed excellent inter-clinician reliability (ICC[2,1] range=0.79-0.97; SEM=1.7°-3.0°) for all measures except nondominant IR which had good reliability(0.72). Conclusions: Results suggest that clinicians can achieve reliable measures of GH rotation and GIRD using a single-clinician technique and an inexpensive, readily available mechanical inclinometer.