dc.description.abstract | With the various development of technology, there is getting more people use their digital or 3C products, which can enlarge and enrich our live. For instance, people used to take a photo with range finder camera but now it is popular for using digital camera or digital single lens reflex camera, even using smart phone. Unfortunately, it is hard to fit the image in different kinds of devices with various ratios. Take the iPhone with screen ratio for example, the height and width ratio is 3:2; people cannot put the image such as 1:1,640*480 those kinds of image into iPhone without distortion in visualization. The solution is to resize the image to the target ratio.
There are several kinds of method to resize the image such as Scaling, Cropping, or Warping. Scaling and Cropping are the fast method to achieve such goal. But the previous one may distort the important object. Although cropping is the fast way to crop the interest area that the user wants, it may lose the other information too. Warping is another way to resize image, but it is not so effective as well.
There is an effective and novel approach called seam carving that could achieve content-aware image resizing. Users may reduce or expand the height and width by themselves. But it may distort the line detection and object shape easily. In this thesis, we focus our research on improving the all line and shape structure in an image, trying to avoid seam goes through that kinds of things. Moreover, we discuss the seam carving on video resizing, which is more complicated than image resizing. At last, the experiments are carried out on different image/video frame with different number of seams. The results demonstrate the prominence of our method. | en_US |