| 摘要 |
A signature capture device includes a CPU that executes a routine to losslessly compress signature data in realtime such that decompression recovers a legally acceptable replica of the original signature. Digital data representing the pixel start point of a signature run is stored in preferably 10-bit or 9-bit format, and is then reduced to a common lower format of 9-bit or 8-bits. Subsequent data points within a run are reduced to the same common lower format and are compared to the common lower format version of the start point data. The differential between the start point and the adjacent second point is calculated and encoded with 4-bits if possible, or with 8-bits otherwise. The differential between the third point and second point is calculated and encoded with 4-bits or 8-bits, as required, and so on, within the run. A run is initiated by a pen-down status, or if a present run exceeds a predetermined length, or if a differential can no longer be encoded with 4-bits and requires 8-bit encoding. For each run, the start position and differential pixel positions are encoded with a preferably 32-bit start block the provides pen-up and pen-down status (1-bit), run length (10 bit), encoding 4-bit or 8-bit status (1-bit) and up to 10 bits each for initial x and y pixel position. The routine is stored in device ROM or as firmware, and incoming data is CPU processed and stored in RAM for export as a compressed data stream.
|
