| 摘要 |
A method and apparatus for recognizing characters in pixel image data and for forming a text file of the characters. Pixel image data is inputted and, if the pixel image data is not binary image data then the pixel image data is converted into binary pixel image data. Blocks of pixel image data are selected by outlining contours of connected components in the pixel image data, determining whether the outlined connected components include text unit or non-text units based on the size of the outlined connected components, selectively connecting text units widthwisely to form text lines based on proximity of adjacent text units, and selectively connecting text lines vertically to form text blocks based on proximity of adjacent text lines and on the position of non-text units between text lines. A hierarchical tree is formed based on the outlined connected components. Text blocks are segmented into text lines of pixel data by adaptively dividing the text blocks into at least one column based on a horizontal projection of pixel density across the column, and characters are cut from the segmented lines in two cutting steps in which the first cutting step cuts between non-touching and non-overlapping characters and the second cutting step cuts between touching characters. The cut characters are recognized and character codes are derived based on such recognition. The character codes are stored in a computer text file in accordance with the order established by the hierarchical tree. If desired, the non-text units may be interspersed with the stored character codes in accordance with the order established by the hierarchical tree. The pixel image data may be pre-processed by, for example, image compression or image enhancement, and the recognized characters may be subjected to post-processing, for example, context checking. Designators may be appended to non-text units based on characteristics of the non-text units. For example, white contour tracing may be employed on the interior of non-text units, non-grid-arranged white contours may be recombined, and the fill rate of the white contours may be calculated, and table designators appended to the non-text unit based on the number of white contours or the recombination rate of non-grid-arranged white contours or the fill rate of white contours. Cutting between non-touching and non-overlapping characters may be accomplished by sparsely stepping through the line segment, and cutting between touching characters may be accomplished in accordance with whether information concerning the spacing between touching characters is known. If information concerning the spacing is known, then cutting may be accomplished based on the spacing statistics. If information is not known, then cutting between touching characters may be accomplished in accordance with rotated projections of pixel density so as to make an oblique cut through the touching characters at an angle and position determined by the rotated projection. Inadvertently characters may be recombined. <IMAGE>
|
