By Dustin Niles
While this blog series has focused on digital images, there is still power in the images that haven’t been taken digitally. Those images are often preserved in prints, which degrade, fade, and otherwise become damaged. One team set out to use the power of the internet to repair images that never made it there. “There is a strong public demand to repair photographs, for example to remove an unwanted object or person from a wedding or travel photo, or to filling missing areas in an old, damaged photo. Image completion provides an effective tool for this purpose. It fills in missing or unwanted regions with new, plausible content.” (Zhu, Huang, Tan, Xu, & Hu, 2015)
The tool works by taking advantage of the ubiquity of internet images. By using the thousands of images of world landmarks on the internet, the researchers were able to find matches to images of world landmarks that needed repair. Using algorithms they could find the closest matches, and then use image editing software to warp the internet images into filling in the missing or unwanted areas of the images. (Zhu, Huang, Tan, Xu, & Hu, 2015)

The researchers' tool would be able to use other internet images of famous landmarks to remove objects like the obtrusive light post in this image.

Photo by Jean-Christophe Benoist. Licensed under Creative Commons Attribution 3.0 Unported.

According to the researchers, the tool showed promise in completing a wide range of images, only running into significant problems if the angle between the user image and the candidate image different greatly, or if the lighting differences were particularly notable. (Zhu, Huang, Tan, Xu, & Hu, 2015)
Regardless, the tool shows the positive power inherent in the deluge of internet imagery that we see every day. While image editing might be commonly associated with negative outcomes and attempts to deceive and misinform, projects like these remind us that not every image on the internet is a negative one, and that we can use the abilities afforded to us by digital media and image editing tools to enhance our own lives and the lives of those around us.

Zhu, Z., Huang, H. Z., Tan, Z. P., Xu, K., & Hu, S. M. (2015, Sep. 18) Faithful completion of images of scenic landmarks using internet images. IEEE Transactions on Visualization and Computer Graphics, 22(8), 1945-1958. Retrieved from stamp/stamp.jsp?tp=&arnumber=7272134&isnumber=7501980
Back to Top