Seam carving based on dynamic energy regulation

Seam carving algorithm is widely used in content-based image scaling. By calculating the energy map of the image, it repeatedly removes the pixel line with the lowest energy sum, which can effectively retain the proportion of significant areas within the image after the image is scaled down. The traditional seam carving does not take into account the variation in full-image energy caused by each carving, which is based on the energy map calculated at the first time. The results of these methods are prone to distortion. So we put forward a dynamic energy regulation method to simulate the energy change in each carving to improve the effect of seam carving. Our method adjusts the energy value of each pixel after each carving according to how much each pixel is affected by carving, so as to simulate the extra energy introduced by each carving. In the paper, we discuss the way to regulate energy. We designed a randomized double-blind experiment to compare our method with several current typical methods. The experimental results demonstrated the advantages of our method over other methods.