The new technique, from the Korea Advanced Institute of Science and Technology, tracks the rotation sensor in a smartphone to improve document rotation. This deals with a problem that arises when users take document photographs using smartphones. Here the images become parallel to the documents on a flat surface, sometimes causing problems when they are rotated.
If you try to take photographs of something like a poster or a brochure and then view the images, it can sometimes occur that the images documents have the wrong orientation (such as rotating 90 degrees). If you try and compensate by rotating your smartphone 90 degrees in the opposition direction, the document photos can sometimes also rotate at the same time, leading to the image always being the wrong-way-round for the user. It is such orientation errors that the South Korean researchers have been looking into.
Heading up the research is Professor Uichin Lee, and he has found the key reasons for these orientation errors. The error is due to a common software glitch in screen rotation-tracking algorithms found with most commercial smartphones. The recent research found, for example, landscape document photos have an error rate of 93 percent across common smartphones.
Professor Lee explains: “When taking a photo of a document, your smartphone generally becomes parallel to the flat surface…Your phone fails to track the orientation if you make any rotation changes at that moment.”
This happens due to software engineers designing the rotation tracking software in smartphones with an important assumption: that people will hold their phones vertically either in portrait or landscape orientations.
This can be overcome, however, by achieving orientation tracking through the measurement of the gravity direction using an acceleration sensor in the phone (for example, whether gravity falls into the portrait or landscape direction). To test this out, Professor Lee ran some studies. These were based on a technique which will correct a phone’s orientation. This is by tracking the rotation sensor in a phone. For this to work it had to overcome a common aspect of photo taking using a smartphone since most people tilt the phone towards themselves (this is called a “micro-tilt phenomenon”). So it was important that smartphone software compensated for patterns of gravity distributions across the device.
By using new algorithms the researchers managed to accurately track smartphone orientation in document-capturing tasks at over 95 percent accuracy. This proved workable on both Google Android and Apple iPhones. It is hoped the new algorithm becomes commercialized.
The research is published in International Journal of Human-Computer Studies and it is titled “Understanding mobile document capture and correcting orientation errors.”