Seam carving

Seam carving, also called liquid rescaling, is an algorithm for content-aware image resizing developed by Shai Avidan of Mitsubishi Electric Research Laboratories and Ariel Shamir of the Interdisciplinary Center and MERL. Instead of scaling or cropping a picture uniformly, it finds seams, which are connected paths of pixels that matter least to the image, and removes them to shrink the picture or duplicates them to enlarge it. A vertical seam holds one pixel in each row from top to bottom and a horizontal seam runs left to right, with importance judged by how strongly a pixel contrasts with its neighbours. The cheapest seam can be found with dynamic programming, Dijkstra's algorithm, a greedy search or graph cuts. Users can also mark regions that must not change, and whole objects can be removed from a photograph. The method addresses image retargeting for screens of different sizes, and Adobe licensed it for the Content Aware Scaling feature in Photoshop CS4.