Image Reconstruction

Renxuan Pang

Introduction

Throughout the development of human history, mankind faces one of their greatest enemies, time. Time acts like a double edged sword, allowing society to grow and develop, on the other hand it also takes away memories from humans, and photos are one of them. As time goes on, photos start to lose their shape, color, or even completely break apart. This is due to the fact that old photos only had physical copies, and there is no access to cloud storage for photos like there are today. One good thing is, the development of artificial intelligence greatly conquered that issue, there are robots that can learn the composition of the photo and restore it to its former glory. Ultimately the things lost in the race of time had been recovered with the power of A.I, and is the first step towards preserving and recovering the ancient human society and culture.

What is the AI?

In the June of 2020, the genealogy website myheritage.com introduced a program that allows “one click fix” of historical photos to either give them color, make them look brand new or even both. The program reconstructs the photo and smooth it out to make the image look like it had been taken by a modern camera instead of the box 20 years ago that only spits out black and white. Despite the results sometimes coming out too smooth to be realistic, people on social media still absolutely loved the fact that it allowed them to see what their grandparents looked like 20 years ago. “Have to admit this new photo enhancer tool by @MyHeritage is amazing! Just discovered my Irish great-grandfather looked like @jeremycorbyn. Maybe we’re distantly related." said Katie Morgan on june 14, 2020, this shows how people had been using this tool to enhance photos taken decades ago to find out what people and towns during those times looked like as well as allowing them to find more details that might truly astonish them or perhaps even change the views they’ve had for their whole life.

How does it work?

First, the program is shown thousands or even tens of thousands of images from everywhere around the world. From the perspective of professionals, “The ‘training’ involves analyzing thousands of normal color photos, as well as black and white versions of the same images, to help it understand which colors correlate to different shapes and textures, explained Breslin, who specializes in engineering and computer science.”(Breslin AI photo restoration shines a light on life in old Ireland cnn.com) Then, it starts to see patterns, such as what type of buildings usually include what colors, what type of color are on which types of clothes, and even the colors of certain trees. Afterwards, the AI would separate the patterns into separate groups, for example it would have a list of colors seen on trees, and whenever the AI sees a tree it will immediately reach for the said group. This is done in order to be faster and more efficient, to prevent it having to go through, and possible overload on tons of data at once. Lastly, the designers would tweak the data a bit to match certain cultures, for example "An average kind of color (of a roof), around the world, might be a terracotta, an orange or a kind of a brownish-type tile...Whereas in Ireland, the roofs were typically slate, which is gray or black." (Breslin AI photo restoration shines a light on life in old Ireland cnn.com) and the very basic idea of an image restorative AI is complete.

How else is it used?

Despite the AI being open to most only in recent years, the idea of processing images via computer science actually came about quite a while ago. Previously, similar AI were actually used in multiple different fields. According to Prateek Majumder, who is a final year engineering student from institution of engineering and management in Kolkata, “In the medical field, Image Processing is used for various tasks like PET scan, X-Ray Imaging, Medical CT, UV imaging, Cancer Cell Image processing, and much more. The introduction of Image Processing to the medical technology field has greatly improved the diagnostics process.” The introduction of image processing and reconstruction AI can greatly speed up the process of diagnosing medical conditions. Another way it is used is for the development of autopiloting cars. “One of the most interesting and useful applications of Image Processing is in Computer Vision. Computer Vision is used to make the computer see, identify things, and process the whole environment as a whole. An important use of Computer Vision is Self Driving cars, Drones etc. CV helps in obstacle detection, path recognition, and understanding the environment” (Prateek Majumber). The image processing lets programs determine and understand the environment around it in real time, which can chain on to certain commands during certain events, allowing for the ability to autopilot machines. Ultimately, the image processing AI, which is the main technology used for the image reconstruction AIs, is convenient in general and allows for many developments in society that we cannot even imagine today.

Conclusion

As the battle against “time” goes on, we will never know what other ways people will come up with in order to preserve the beauty of history. As far as time goes, the image reconstruction AI is a great leap toward that goal, and is one big puzzle piece leading to the discovery of all the bits of time. In the future, we might be able to see all of history recovered in color with technological advancements, and the image construction AI would be one of the building blocks. Unlike how hard it is to predict the future, history remains within our grasp. The step has been taken, and it shall lead to something so much bigger, something that only what humans have been fighting, time, will tell.

Citations

  1. Cade, Dl. “The MyHeritage ’Photo Enhancer’ Uses AI to Restore Old Family Photos.” PetaPixel, 23 July 2020, https://petapixel.com/2020/07/23/myheritage-unveils-photo-enhancer-tool-that-uses-ai-to-restore-old-photos/. Accessed 5 December 2021.

  2. Charlesworth, JJ. “AI Can Produce Pictures, but Can It Create Art for Itself?” CNN, Cable News Network, 6 Sept. 2018, https://www.cnn.com/style/article/artificial-intelligence-ai-art/index.html.

  3. Holland, Oscar. “AI photo restoration shines a light on life in old Ireland.” CNN, 29 March 2021, https://www.cnn.com/style/article/old-ireland-color-photos-breslin/index.html. Accessed 5 December 2021.

  4. MAJUMDER, PRATEEK. “Digital Image Processing Applications | Image Processing in Python.” Analytics Vidhya, 28 May 2021, https://www.analyticsvidhya.com/blog/2021/05/digital-image-processing-real-life-applications-and-getting-started-in-python/. Accessed 5 December 2021.