| 摘要 |
PURPOSE:To convert plural lane videos to primary information for every lane simultaneously, by using a reference element storage table, where boundary posiions are stored, to convert an input pattern expressed as the time series of vector to a scalar quantity. CONSTITUTION:A timing generating part 15 supplies the synchronizing signal to a control part 14 synchronously with the picture scanning timing. The control part 14 refers to region boundaries stored in a reference element storage part 16 to switch a gate 10 at a scanning timing corresponding to boundary positions of lanes on the scanning line and supplies a partial vector of vector to an operating part 11. The operating part 11 converts the partial vector to a scalar quantity. A gate 12 is switched at the timing opposite to the gate 10 to take out the operation result. A changeover switch 13 takes out one-dimensional vehicle information corresponding to every lane. Thus, plural lane videos are converted to primary information for every lane simultaneously in real time with a high precision cheaply. |
