In other swipe livescanners, the images are taken apart some time and each one represents part of the total image with some overlap.
See for instance this paper, where one Atmel chip, similar to the Egistec one, is used: http://3dvision.joanneum.at/3DVision/publications-presentations/literature/pdfs/PUB04DIB015.pdf They used a mathematical adjustment to get the registration of the different images of finger as it swipes in the main movement direction but at same time can slide in the lateral direction, to the right or to the left. See fig 3 on paper. The superposition between the images depends upon the swipe speed, so you can get some black lines if you swipe too fast. I think that with the 8 images you can use any graphics software to find the superposition of subsequent images. Unless you swipe very steadily and with constant speed, the superposition will vary between subsequent images. To do it automatically, you can try some mathematical adjustments or the procedure described in the paper. Any way, good work! A C Censi Any way, it is nice that you could get an image without encryption, because this chip can encrypt the data transmitted. See datasheet from Egistec. 2011/2/8 Алексей Прокопчук <alex...@homelan.lg.ua>: > I did it! :) > I finally realized how get the image from the received data. > Perhaps, the explanation will turn chaotic, but I'll try to explain as > precisely as possible. > On buffer with 61440 bytes length stored an image 192x80. And that's very > interesting that the correct pixel is located in the lower four bits of > every fourth byte. Moreover, any fourth byte. For example: > I can save PGM image with max gray level 15 from 61440 buffer by four ways: > > char buf[61440]; > > buf[0]&0x0F, buf[4]&0x0F, buf[8]&0x0F ... > buf[1]&0x0F, buf[5]&0x0F, buf[9]&0x0F ... > buf[2]&0x0F, buf[6]&0x0F, buf[10]&0x0F ... > buf[3]&0x0F, buf[7]&0x0F, buf[11]&0x0F ... > > As a result, are four almost identical images, visual difference is almost > insignificant. On the images very clearly visible the fingerprint. > Although it seems to me to invert the colors, so it will be correct: the > black lines of the fingerprint on a white background. In the initial > buffer all the way around, the white lines on black background. > > Now when understandable how to convert a raw buffer to the image, another > questions arises ... All eight images - a sequence of sweeping finger on > the working surface of the scanner. And now it is unclear how to combine > the eight images into one that contains a full fingerprint. And what > resolution get this picture as a result... And in which format should I > send data to the processing libfprint after final assembly of the image? > > Thanks in advance! > > > _______________________________________________ > fprint mailing list > fprint@reactivated.net > http://lists.reactivated.net/mailman/listinfo/fprint > -- A. C. Censi accensi [em] gmail [ponto] com accensi [em] montreal [ponto] com [ponto] br _______________________________________________ fprint mailing list fprint@reactivated.net http://lists.reactivated.net/mailman/listinfo/fprint
