Let's face it, it is extremely difficult to upconvert Standard Definition (SD) television images into high quality High Definition (HD) images. It involves deep image processing, mathematical and real time processing problems. Yet, upconversion is now a central issue for the broadcast and consumer electronics industries.
Multiple formats and format conversions
High Definition broadcast channels need to convert existing Standard Definition archives, films, advertising into an HD format to broadcast them over their HD channel, while maintaining high quality images.
Moreover, HD flat LCD or plasma screens and video projectors must display progressive scan HD images whatever the format of the incoming source, which is most often interlaced standard television images.
The number of video formats is increasing beyond reason, with various flavors corresponding to different image sizes (number of rows and columns), different time sampling rate (number of images per second) and different space-time sampling pattern (interlacing or not).
Figure 1: Sizes in rows x columns of Standard Definition and High Definition television formats
Upconversion of interlaced videos
Getting high quality HD images from SD contents requires a super-resolution process that computes missing pixels from available ones, with the best possible resolution. One can see it as a two-step process: deinterlacing, which computes the missing even or odd rows of each field of an input interlaced video and scaling to increase the image size from input to output resolution. The output image quality depends mostly upon the efficiency of deinterlacing, which is the most difficult process.
Video/film cadence
Interlaced videos often have a more complex time structure. This happens when films are converted into interlaced videos or when elements such as crawling text or logos are inserted in images. Amongst these, the famous 3:2 pulldown and 2:2 pulldown cadences are produced by repeating fields in time to convert film 24Hz input to 60 or 50Hz video frequencies. With the diversification of video sources the variety of cadences gets more and more complex. A mix of these cadences may occur at the post-production stage and per pixel cadence detection is then require to properly handle these kinds of contents.
Deinterlacing and scaling
In the 1980's and 1990's, the first challenge of deinterlacing was real-time digital video processing at a minimum cost. Minimizing the number of operations and memory requirements was thus necessary, which first led to two simple methods: spatial line doubling and time weaving.
Next: Motion adaptive deinterlacing
|
