To provide an image matching method that reduces the amount of calculations for a dynamic equation.
Target grid points are set in a target image and reference grid points are set in a reference image in a one-to-one relationship with the target grid points. The target image is divided into clusters and the typical target grid point of each cluster is determined. Based on the pixel information of a typical reference grid point serving as a reference grid point (y) corresponding to the typical target grid point, the pixel information of each typical target grid point, and the gradient of image correlation potential energy determined by the positional relationship between the typical reference grid point and each typical target grid point, the image correlation force that the typical reference grid point receives is calculated. The image correlation force acting on the reference grid point (y) corresponding to each target grid point (x) belonging to each cluster is set as the calculated image correlation force. Using the image correlation force acting on each reference grid point, an elastic force that each reference grid point receives because of elastic energy between each reference grid point and the adjacent reference grid point, and a friction force acting on each reference grid point, a dynamic equation about the reference grid points is constructed. The dynamic equation is numerically analyzed to determine the equilibrium state of each reference grid point (y).
ITO TAKESHI
BABA MASAHIRO
Satoshi Kono
Makoto Nakamura
Kurata Masatoshi
Takashi Mine
Yoshihiro Fukuhara
Sadao Muramatsu
Ryo Hashimoto