Seam carving is an algorithm that performs content-aware rescaling on image and videos. Content-aware re-scaling takes into account important features within the frame, and intelligently targets parts of the frame to remove during the re-scaling process. This process distorts the image less than other resizing approaches such as cropping or scaling.
This algorithm is computationally complex and therefore usually used for images rather than videos. Our goal is to use a hardware-oriented implementation of seam carving on a FPGA to improve performance in order to resize videos. We will also modify the algorithm to take into account how the important features in the video move over time so they are not removed.