Development of saliency‐based seamless image compositing using hybrid blending (SSICHB)

With the advancement in computer vision and graphics tools, digital compositing has become an integral part of the present computer‐generated visual effects. However, factors such as inaccurate mask generation, intensive user interaction, and the creation of boundary seams due to colour or texture difference, make it hard to achieve quality composites. Poisson editing efficiently generates seamless composites but results in undesirable colour bleeding. Multi‐resolution blending, in contrast, produces colour consistent composites; however often introduces blurry boundaries around the source object inserted. Motivated by these observations, the authors propose a colour consistent seamless compositing pipeline by integrating two new approaches. First, the authors use a visual attention algorithm based on the colour difference with increased edge weight using Gaussian filter bank (GFB) to ensure the least user interaction during mask generation. Second, the authors propose a hybrid framework by incorporating the goodness of the two different blending methods namely modified Poisson editing (MPE) and GFB‐based multi‐resolution blending to create colour consistent seamless composites. An extensive experiment has been carried out on challenging datasets to validate the proposed technique. Comparison with the state‐of‐the‐art techniques shows the efficacy of the authors’ algorithm in generating colour consistent, seamless, and natural‐looking composites for present image editing applications.