Interactive image resizing using seam carving and object detection

摘要

Seam carving is an image processing operation for content-aware image resizing. However, it sometimes fails to protect important objects in the image. In this paper, an interactive image resizing technique is proposed by combining seam carving and object detection to detect the object (region of interest) based on a single Gaussian background model. Compared with the inter-frame difference method, the single Gaussian background model is more suitable for object detection because of its better accuracy. After object detection, it is convenient for selecting the objects to be retained or removed in the image. Experimental results demonstrate that the proposed approach can easily remove or protect the region of interest in an image and image resizing is achieved.