Our client, a California based, post-production company in its movie restoration projects uses a highly complex technology to scan the footage and automatically improve the quality of each movie frame in an intense movie restoration procedure that generates all the frames in full HD format.
Their need resided in matching the full HD frames to the low quality frames of the original movie.
We developed a software solution using advanced matching algorithms to find the correspondence between the full HD pictures and the original video frames that were used as input. The level of accuracy for software is very high, providing three states of confidence (high, medium, low) for the probability of the match and also gives the user the possibility of manually assessing some of the frames. Based on these correspondences the movie is recreated in Full HD format.
In order to obtain the matching between frames and pictures as fast as possible, we’ve implemented data structures for spatial search and a Vantage Point Tree algorithm that skips some pre-processing steps and saves time.
The back end of the application was developed using C++. For implementing the front-end part we have used .NET technologies (Windows Presentation Foundation). The project was developed as a desktop application and it is used by post-production technicians