Content-aware Image Resizing

As the development of various display devices and internet applications, automatically resizing images to fit the display target is becoming more important. Content-aware image resizing technique which protects prominent information in the image has been an active research topic in recent years.We summarize the recently developed content-aware image/video resizing method by introducing their technique characteristics and analyzing their advantages and disadvantages. After that, we present a new image resizing method based on seam carving method.At first, our method combines the image gradient and the visual saliency to a balance energy map to measure the cost of the seams. Then it proposes a method to evaluate the diagonal artifacts in addition to the previous horizontal and vertical artifacts for the forward seam carving method. At last, we develop a simple high-level saliency detection method based on rule of thirds of photo composition knowledge to constrain the seam carving procedure for protecting the foreground contents.The experimental results showed that the proposed method can improve the visual quality of the re-targeted image and the robustness of the seam carving method. Moreover, the improved method is simple to implement, and can be easily applied to many existing image resizing methods.