Soil computer tomographic (CT) images sequence are difficult to extract the target because the gray of the segmented object has no uniform distribution. In order to solve the problems, we come up with an adaptive segmentation algorithm of CT images sequence based on structure continuity as prior information. We choose the initial image and determine the most appropriate threshold. Based on the continuity of the same object structure between the images sequence, the areal change of the object region is used as the prior information to perform recursive segmentation. The adaptive segmentation of CT images are achieved, and 3D visualization is realized by 3D reconstruction on the segmented object area. The results show that the proposed algorithm is applicable to CT images sequence with non-uniform gray distribution in the object regions, it results in small amount of calculation, and can precisely segment objects in thousands of target images.