| 摘要 |
PURPOSE:To attain high precise recognition by executing a feature point detection processing by a hardware synchronizing with a clock signal, and identifying whether or not a picture element position is an inflection point by the combination of the compared results of a density value with a density threshold value, and a density changing amount with a differential threshold value. CONSTITUTION:Each constituting element of an arithmetic part 154 is described at a size and position matched with a clock timing so as to be matched with the timing of an hourly delay and processing. Comparators 261 and 262 compare a density differential value DIFF outputted by a computing element 232 with threshold values ADIN and ADIP, and identify whether or not the direction and changing amount of a density change are sufficiently large. Moreover, comparators 271 and 272 compare the density differential values of the picture element position, and respectively generate two signals. Then, the changing shape of the density differential value DIFF is checked based on those signals, and the top part of the mountain and the bottom part of the valley of the change of the value DIFF are detected as an outline position. |
