An image of a physical object is produced by receiving a plurality of raw images, dividing the plurality of raw images into a first subset of primary images and a second subset of secondary images according to a predetermined criterion. From the first subset of primary images an intermediate image is determined while from the second subset of secondary images a mask image is determined. Afterwards a registration of the intermediate image and the mask image is performed by using direct registration of predetermined features present in the intermediate image and the mask image. A fused image of the physical object is generated out of the mask image and the intermediate image.