1992 年 441 巻 p. 103-113
Remarkable progressing on technical computer image analysis shows us inexperienced but possible method to grasp man's visual recognition quantatiyely from digital image. Two characters, horizontality and verticality, seem to strongly concern with optical perception of street view. In this paper, we explained the concept of measure to grasp strength of direction, which is defined by edges and lines got from digital image, and proposed measures for horizontality and verticality. As the result of apllication of measures to 64 images of street veiw : (1) We made possibility clear to grasp strength of horizontality and verticality. (2) We got 4 comprehensible clusters by cluster analysis, using 3 quantative measures proposed in this paper.