Thesis (Ph. D.)--Ohio University, March, 1995. / Title from PDF t.p.

Thesis (M.S.)--Ohio University, August, 1992. / Title from PDF t.p.

Thesis (M.S.)--Ohio University, November, 1992. / Title from PDF t.p.

Thesis (M. Phil.)--University of Hong Kong, 1992.

Thesis (M. Phil.)--University of Hong Kong, 1998. / Includes bibliographical references.

Thesis (Ph. D.)--University of Washington, 1996. / Vita. Includes bibliographical references (leaves [168]-173).

-&gt;Context. Optical Character Recognition (OCR), having a computer recognize text from an image, is not as intuitive as human recognition. Even small (to human eyes) degradations can thwart the OCR result. The problem is that random unknown degradations are unavoidable in a real-world setting. -&gt;Objectives. The noise tolerance of Tesseract, a state-of-the-art OCR engine, is evaluated in relation to how well it handles salt and pepper noise, a type of image degradation. Noise tolerance is measured as the percentage of aberrant pixels when comparing two images (one with noise and the other without noise). -&gt;Methods. A novel systematic approach for finding the noise tolerance of an OCR engine is presented. A simulated environment is developed, where the test parameters, called test cases (font, font size, text string), can be modified. The simulation program creates a text string image (white background, black text), degrades it iteratively using salt and pepper noise, and lets Tesseract perform OCR on it, in each iteration. The iteration process is stopped when the comparison between the image text string and the OCR result of Tesseract mismatches. -&gt;Results. Simulation results are given as changed pixels percentage (noise tolerance) between the clean text string image and the text string image the degradation iteration before Tesseract OCR failed to recognize all characters in the text string image. The results include 14400 test cases: 4 fonts (Arial, Calibri, Courier and Georgia), 100 font sizes (1-100) and 36 different strings (4*100*36=14400), resulting in about 1.8 million OCR attempts performed by Tesseract. -&gt;Conclusions. The noise tolerance depended on the test parameters. Font sizes smaller than 7 were not recognized at all, even without noise applied. The font size interval 13-22 was the peak performance interval, i.e. the font size interval that had the highest noise tolerance, except for the only monospaced font tested, Courier, which had lower noise tolerance in the peak performance interval. The noise tolerance trend for the font size interval 22-100 was that the noise tolerance decreased for larger font sizes. The noise tolerance of Tesseract as a whole, given the experiment results, was circa 6.21 %, i.e. if 6.21 % of the pixel in the image has changed Tesseract can still recognize all text in the image. / <p>42</p>

The subject of machine recognition of connected characters is investigated. A generic single character recognizer (SCR) assumes there is only one character in the image. The goal of this project is to design a connected character segmentation algorithm (CCSA) without the above assumption. The newly designed CCSA will make use of a readily available SCR. The input image (e.g. a word with touching letters) is first transformed (thinned) into its skeletal form. The CCSA will then extract the image features (nodes and branches) and store them in a hierarchical form. The hierarchy stems from the left-to-right rule of writing of the English language. The CCSA will first attempt to recognize the first letter. When this is done, the first letter is deleted and the algorithm repeats. After extracting the image features, the CCSA starts to create a set of test images from the beginning of the word (i.e. beginning of the description). Each test image contains one more feature than its predecessor. The number of test images in the set is constrained by a predetermined fixed width or a fixed total number of features. The SCR is then called to examine each test image. The recognizable test image(s) in the set are extracted. Let each recognizable test image be denoted by C₁. For each C₁, a string of letters C₂, C₃, CL is formed. C₂ is the best recognized test image in a set of test images created after the deletion of C₁ from the beginning of the current word. C₃ through CL are created by the same method. All such strings are examined to determine which string contains the best recognized C₁. Experimental results on test images with two characters yield a recognition rate of 72.66%. Examples with more than two characters are also shown. Furthermore, the experimental results suggested that topologically simple test images can be more difficult to recognize than those which are topologically more complex. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

Master of Science

No description available.