A novel technique for mosaicing of medical images

摘要

Image mosaicing has numerous applications in the area of computer vision, virtual reality, robotics, medical imaging, computer graphics and advertising. The creation of the seamless image mosaic often becomes a difficult task due to the presence of alignment errors and intensity discrepancies. The present paper discusses the development and implementation of a novel algorithm based on the discrete cosine transform (DCT) for mosaicing of medical images. The proposed method is implemented in three steps: image alignment, image reprojection, and image blending. In the first step, the geometric correspondence between the images to be mosaiced is calculated. The aim of the second step is to project the aligned images in a common compositing surface. Finally, the third step minimizes the intensity variations in the overlapping region of the final mosaic. As a result, a high quality seamless image mosaic is generated. The obtained results and its comparative analyses clearly illustrate the efficacy of the approach in obtaining visually pleasant mosaiced images.