In view of the problems of large number of original data layers in computed tomography (CT) image acquisition and ease errors in the reconstruction process, we propose an improved FBP algorithm, which is integrated with Amira software for the 3D reconstruction of CT images. First, the improved Ram?Lak filtering is applied to CT images using MATLAB software for filtering. Second, the CT images are also binarized and cropped for preprocessing, the improved FBP algorithm is used to reconstruct two-dimensional CT images and generate multiple tomographic images. A comparison is made between the reconstruction results of the traditional FBP algorithm and the improved FBP algorithm. Finally, Amira software is employed to perform a three-dimensional reconstruction of the multiple tomographic images. The experimental results show that the proposed method can effectively solve problems such as redundant data collection, large spatial crosstalk, poor imaging effect, and high radiation dose in the actual clinical CT imaging process. Moreover, the obtained tomographic image edges are clearer, and the reconstructed 3D image of the finger is more intuitive and stereoscopic.