Budget Amount *help |
¥2,750,000 (Direct Cost: ¥2,600,000、Indirect Cost: ¥150,000)
Fiscal Year 2007: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2006: ¥2,100,000 (Direct Cost: ¥2,100,000)
|
Research Abstract |
When the characters are written with pen, by writing pressure, the paper was indented locally. ESDA2 is the equipment for detecting the indented pattern. The patterns translated with ESDA2 are black on the indented areas by adsorbing the toner: However, this equipment detects the indented patterns throughout some sheets of paper and makes a sheet of multiplex indented pattern because of its high sensitivity. In this research, to separate single indented pattern from the multiplex pattern was examined for interpreting the written characters. First, the original images of the upper and the lower paper were binarized by gray-levels. Second, the black pixels being at the low density of the neighbor black pixels were removed. Next, dilation and erosion were executed to reduce the isolated black pixels. Finally, subtraction between both images was executed. If the pixel pl was white(0) and the pixel p2 was black(1), pl-p2(-1) was converted to 0(white). Thus, the black pixels in the image to be subtracted were remained. It was possible to interpret the separated pattern, though the image quality was not so good. In the traditional method of photographing the shadows of the edges of the indented patterns with oblique lighting, it is impossible to take the indented patterns parallel to the light axis. Therefore, previously, it was impossible to visualize the shapes of the characters consisting of various directional strokes completely. In this research, the images which were taken individually with some directional oblique lighting were superimposed by multiplication. As a result, the character patterns which are interpreted easily are synthesized on an image. The unevenness of gray-levels on the superimposed image due to the deflection of the paper or the spreading of light was reduced drastically by the processing of making even the average gray-levels in the local areas on the original images.
|