Project.2

#1

In this picture, I have decent results for both vertical and horizontal seam carving. I think this picture worked fine because the main object (whale) is well distinguished from the background (sea, mountain, sky, and clouds).

#2.

    This example looks like it's cropped. However, if you look at the right side edge, the sand still hit the ground. If I cropped the image, that wouldn't happen. (I forgot where I took this image from. on the Internet??)

#3.

    This example gives me some interesting result. A cup, which looks almost perfect circular shape, became a funny shape. We can realize that upper part for the horizontal seam craving, and left side for the vertical seam carving is changed more than other parts. I guess the reason is that when we call min function, if there's same minimum value, it just gives us first minimum value and since we originally have a circle shape, we probably have a same minimum values on upper and bottom parts, left and right side at the same time. However, since our min function returns the first minimum values, it gives me this weird shape. These weird shapes give us an idea how the original image is transformed.

#4.

    This picture is just for the horizontal seam craving. As we can see, there's no interesting going on the vertical seam carving. I compared horizontal seam carved one with original version and it exactly matched to lower part of orignal image. I guess my function tends to save green more than gray (sky).

#5.

    On this example, I got interesting result on horizontal seam carving one. As we observed on #4, because of the tendency to preserve green area more than others, my function think face of the baby has less important than others. on the other hands, vertical seam carving worked fine.   (This is my cousin.)

#6.

    On this example, both vertical and horizontal seam carving worked perfectly fine. It's interesting to see a reflection on the water surface is also shrunk as real object shrunk on vertical seam carving. It's also interesting to see the tallest tree, which doesn't have many leaves, is the most shrunk one. Again, it supports my theory that my energy function gives more weight on green color than others. 

    Based on what I got from my function, we can easily think better energy function. When we apply this function to image, we also provide the major color of the object we want to keep and also gives the color which we want to keep the least. (This is a picture of Lake Park in Korea)