Propagating object selection across multiple images

摘要

Propagating object selection across multiple images is described. In one or more embodiments, an object is selected from a first image according to selection inputs. Features that are representative of the selected object are then ascertained. To propagate selection of the object to the second image, a determination is made as to whether the second image includes features that resemble those ascertained from the first image. Once a determination is made as to the features of the second image that resemble the features of the first image, the second image is divided into regions. Each of the regions includes pixels that are similar within the region, one pixel to another. The regions of the second image that include the features resembling those ascertained from the first image are then combined. The combined regions are used to select the object from the second image without receiving selection inputs to do so.

主权项

1. A computer-implemented method comprising:
ascertaining features from a first image that are representative of an object of the first image; determining whether a second image includes one or more features that resemble the features ascertained from the first image; responsive to a determination that a plurality of the features of the second image resemble at least one of the features of the first image, dividing the second image into similar-pixel regions, each of the similar-pixel regions including pixels that are determined to have less than a threshold difference in pixel density with respect to a seed pixel of the similar-pixel region; determining the similar-pixel regions of the second image that include the determined resembled features; and combining the similar-pixel regions of the second image that include the determined resembled features to automatically select the object from the second image.