Improved Image Resizing using Seam Carving and scaling

Seam Carving, the popular content-aware image resizing technique, removes or inserts seams of low energy iteratively without sufficiently considering their impact on the global visual quality of the image. However, sometimes seams pass through the regions of interest (ROIs) with low energy and distort the geometric shapes of the important objects. In this paper, we propose an improved resizing method based on improved seam carving method that can prevent the seams going through the ROIs and better preserve the content of the image. With considering frequency-tuned saliency map and the distance to the image center for each pixel in the importance map, it can protect the ROIs when the image is resized to the target size. Moreover, a developed two-dimension search scheme for seam carving and an switching scheme between seam caving and scaling are also proposed to efficiently protect the global visual of the image. Experiments demonstrate that resizing results of the proposed method are more pleasant than those of cropping, uniform scaling and other several methods.