Are reflections on photos soon gone? Researchers at Massachusetts Institute of technology (MIT) and Google research have apparently developed using an algorithm a method that automatically out targets to reflections from photographs or minimize.
This can be particularly useful if photos–taken up through a window, for example, in the delivery of the transaction. But even then, if a subject on a reflective surface, sometimes annoying reflections occur. The newly developed algorithm should be now noisy with news able to detect double reflections and to remove from the image. They occur about two glazed Windows, but also very thick slices. But this also means that the software covers every case – if there is only a reflection.
Smartphone Cameras And Self Propelled Cars
With the technology, a photo is divided into squares, which are 8 x 8 pixels in size. To identify the reflections on the basis of the distances, the algorithm relies on statistics associated with each image point in connection. For this to work as precisely as possible, he was literally trained with 50,000 test photos.
If the software is further refined, it could be used soon in digital cameras and Smartphones. Also machines that rely on Visual computer-supported, could better estimate so surfaces on which reflections occur. These include, for example, robots in factories and self-propelled cars.
When the software is developed but so far, that she can be used in cameras, is not yet known. Currently, sometimes small artifacts remain even after counting out the reflections back, like the sample images in the publication to the software show.