The creation of mosaics is an interesting application, which leads to many challenging theoretical and practical problems. In this project, we are mainly focusing on viewing the mosaicing as a multiple image alignment task. Therefore, pairwise registration methods with a consecutive normalization and simultaneous registration techniques are in focus. Espcically the simultaneous alignment of images for mosaicing seems very promising. We earlier developed a new type of multivariate extension for similarity measures, which led to a whole new class metrics. They are specifically adapted for their usage in mosaicing, but can also be used for general multiple image alignment tasks, e.g. population studies.
Our main interest lied so far on the combination of several ultrasound images. To this end, we are benifiting from a quite recent development in ultrasonographic research; the development of advanced 2D array ultrasound transducers. Their new fabrication process allows for higher resolution images with a higher quality resulting from the increased frequency spectrum that they can be operated with. The instantaneously acquired ultrasound volumes serve as input for our mosaicing framework, where they will be compounded to one lager volume.