Reliable sources informed that Skype, the video calling service is utilized for near about last ten years. It is informed that more than 300 million users utilize this advanced service and make almost two million video calls every day. However, like most of the service, this one also has certain drawbacks. The users reported that they are unable to look each other in the eye. In spite of the fact that making a video call is a simple process, the caller seems that the other person of the conversation is looking at their chest or over their shoulders.
Claudia Kuster is reported to develop a way-out that she believed would help in to overcome the problem. Claudia, a doctoral student at the Computer Graphics Laboratory ETH Zurich and her well developed team have invented the method to bring eye contact to the Skype and related video services by using advanced software that is proved to change the on-screen image of the callers by providing them the illusion that they are looking straight towards the camera.
Sources informed that to restore the eye contact is quite a difficult and expensive process as of now as it involves using multiple cameras or mirrors, and special types of software. Claudia informed that the technique they have developed used simpler equipment and also operated in real time.
The latest system which is still under development presently used Kinect that serve the purpose of a deep sensor. Claudia stated that the software can be adjusted according to the preference of the users in few simple steps and is also quite robust.
The exclusivity of the technique is not in the process of taking image but in the way, it uses the image of the person. Presently, the entire image is tilted on gaze-correction systems that take a lot of computing power.
Claudia explained that instead of shifting the entire image, their system deals with the foreground. The Kinect generates a depth outline of the image and isolates the person in it along with the border area around it that is usually used to merge the person into the background. After this, the software designs a coordinate mask onto the features of the person and calibrates it. Then the system can tilt the face to the proper angle that seems to appear that eye contact has been made.
In the next step, the face is extracted from the image with the help of a state-of-the-art face tracker. It is informed this would help to compute 66 features points over the face and is also utilized to change the tilted image so that it becomes consistent and is also synchronized with the movement of the person. Finally, the image is pasted back on the image of the person and merged it and this allows matching shadows and hairlines. However, they further informed they are still unable to match the spectacles yet.
Claudia and her team expect to build up the software that would operate with quality cameras not only in the personal computer but also in smartphones and tablets.