| 摘要 |
PURPOSE:To determine a personal feature automatically by digitizing a picture image of the face of a person to be determined, and determining with which feature type among the plural feature types registered previously in a control part the image is matched by means of a neural network. CONSTITUTION:A person to be determined is stood on a photographing stage 14 and his face is photographed by a camera 15 so as to be displayed on a screen of a TV monitor 16 as a digitized contrast gradation picture image and also to be taken into a control part, 12. In the control part 12, one part of a face in the width direction is divided so as to find horizontal histogram inside a partition on the basis of the predetermined characteristic part of the face, and the face characteristic parts such as bows, eyes, nose, and mouth are recognized from the difference in gradation, while the size between the respective parts of the recognized face or the size from each of the parts to the predetermined position of the face are measured, and consequently it is determined which feature type among plural feature types, which are registered previously in the control part, 12 as teacher data, matches with the face by means of a neural network. In this way, it can be determined automatically which feature type matches the face of the person to be determined without depending on any artificial determination. |
