Abstract
Volume data, that can express the inside of the object, is used for an image diagnosis in the medical field and non-destructive testing in the industrial field. By the way, several volume data for the same object, has the deviations of color, translation or rotation caused by the measurement situation. Thus, it is not possible to compare two volume data by superimposing without alignment. However, accurate alignment is not easy. In this paper, we propose a feature quantity of volume data by using D2 feature quantity that is invariant with rotation, and a method of regional search of volume data.