Graph Cut Optimization and Its Application in Image Blending

In the area of computer vision, optimization is an important issue. One of the new developed techniques that caught my attention recently is “Graph-Cut”. I worked on using this new optimization technique in applications such as image blending or image segmentation. I presented a novel approach for combining a set of registered images into a composite mosaic with no visible seams and minimal texture distortion. Pair-wise image blending was performed by means of watershed segmentation and Graph-Cut optimization. This approach is fast because searching the optimal intersection between the images was restricted over a small set of watershed segments, instead of optimizing over the entire set of pixels in the intersection zone. The solution is found efficiently via Graph-Cut, using a given photometric criterion.