High-speed high-accuracy 3D rotation of volume images

3D rotation of volume images has become increasingly important for 3D image analyses, e.g. multi-modality registration and visualization of 3D medical images. A high accuracy of rotation is fundamental, and a high speed of rotation is also significant for many applications. Straightforward extension of the multi-pass scanline methods developed for 2D image rotation does not offer high speed and accuracy. In this paper, we developed a high-speed, high-accuracy method to perform 3D image rotation with size-keeping (i.e. without scaling), by extending the incremental line algorithms (e.g. the mid-point line algorithm). The proposed method is fast since it mainly uses integer addition, increment-by-one and logic justification. The new method is also of high accuracy. The computation and accuracy of the method are given in comparison with those of the multi-pass methods for 3D image rotation.