An Efficient Retargeting Technique to Mitigate Deformations by Integrating Seam Carving and Image Stitching

In the realm of image resizing, the seam carving (SC) method is commonly utilized for altering the image's aspect ratio. The resizing process involves the iterative removal of optimal seam paths from the image. However, researchers have noticed that this removal often results in the image splitting into separate, disjoint segments. The conventional SC approach creates false seams when attempting to merge disjoint segments using the pixel shifting technique. The distortions caused by pixel shifting are quite evident to the human eye. In this paper, a new SC method is proposed that effectively merges the disjoint image segments. The proposed SC integrates proposed image stitching methods to reduce the formation of false seams during the merging process. The proposed image stitching method employs techniques such as feature detection, feature matching, homography estimation, image warping, as well as stitching and blending. The effectiveness of the proposed SC method is assessed through SSIM based on parameters like luminance (I), contrast (c), and structure (s). The objective analysis discloses that the proposed SC method performs well as compared to existing SC. According to the comparison study, the suggested SC yields findings that are 10% more accurate than those obtained using the current method. The suggested SC technique ignores the creation of artificial seams during the image retargeting process while enhancing the image's overall visual quality.