Fast and robust variational optical flow for high-resolution images using SLIC superpixels

We illustrate how a pixel-based method can be directly applied on a near-regular grid of segments (superpixels) from a well-chosen image segmentation. This allows the accelerated processing of high-resolution content with existing pixel-based methods, which might not be able to process it in its input resolution in the first place. In essence, the superpixel segmentation is a tool to decrease the dimensionality of the input image without significant loss of quality for the final result. We show how optical flow estimation, whose execution time is often of vital importance, can benefit from this workflow for processing 4K content whilst also conferring a degree of noise robustness: this pre-processing is a novel direction of research for the field. Naturally, the application of this concept is not restricted to optical flow: it is applicable in many different fields.

from #MedicinebyAlexandrosSfakianakis via xlomafota13 on Inoreader http://ift.tt/1KWgTX2
via IFTTT