摘要 |
PROBLEM TO BE SOLVED: To extract information, which visually represents features of an object, from segments described about the object. SOLUTION: A document is divided on the basis of contents and constitution to generate segments (S301), and object information is referred to determine correspondence information indicating which object a segment corresponds to information of (S302). The segment is analyzed to extract the area of an included image (S303). Object information and correspondence information are referred to count the frequency in appearance of the name of the object corresponding to the segment (S304), and this frequency in appearance and the area of the image are referred to calculate the score of each segment (S305), and this score and correspondence information are referred to determine a representative segment of each object (S306).
|