Deinterlacing

Interlaced video used in television is a sequence of half-frames, each of which carries only half of the visual information (odd half-frames consist of only odd lines, even lines are of even ones). If in each frame we combine the previous half-frame with the current one (for example, even lines from the current half-frame, odd lines from the previous one), jaggedness appears on moving objects: the edges of the objects will look like a comb. This is due to the fact that the phases of the movement are fixed in two half-frames at different points in time, which then when combined will have a mismatch of even and odd lines. To get rid of this unpleasant effect, various mathematical methods are used, which are called deinterlacing.

• The simplest deinterlacing technology is mixing in some proportion an interpolated current half-frame with the previous one. This technology leads to losses of both temporal and spatial resolution, as well as to the "ghost effect" ( eng. Blend ): behind a fast moving object a translucent "ghost" is visible.
• If the video is obtained by re-filming a film using a TV camera or converting a non-interlaced source with a small frame frequency (up to 25 fps in PAL / SECAM , up to 30 in NTSC ), then by guessing the source frequency and the conversion algorithm, you can get an almost exact copy of the original video. This operation is called reverse pulldown or inverse telecine (see telekino ) ..
• Bob deinterlacing. If the video was shot with a TV or video camera (50i), you can decompose interlaced video by separating even lines from odd lines and getting full-fledged 50 frames (in the case of PAL / SECAM and NTC). The resulting video will be compressed vertically and will require stretching to 4: 3 format, which will result in a loss of vertical resolution, but will allow digital-free recording of recordings made in 50i format (concerts, TV shows, as well as video taken with home video cameras, etc.) .)
• Adaptive deinterlacing is a family of algorithms that are determined by a sequence of frames, whether the image is static or dynamic. For pixels that are fixed, the half-frames simply merge without blending. In dynamic pictures, semi-frames are mixed together. Motion compensation can also be applied - for moving objects, deinterlacer tries to fill in the missing information with information from where this object is located on the previous / next frame. Adaptive algorithms provide better image detail, but require more computation. In addition, on compressed interlaced video deinterlacers with motion compensation tend to create non-existent details.

In video surveillance systems , deinterlacing is a marketing move rather than a useful feature. The fact is that during the examination of the video, it is often necessary to examine freeze-frames — for example, to count the number of a stopped car or to obtain an identikit of the person taken. Therefore, at the same bandwidth, the low-frequency progressive-phase scan is preferable to interlaced. If the camera still gives an interlaced stream, recording in an interlaced format is better than deinterlacing.

• Telecine / IVTC Conversion
• Progressive scan in projector and dvd player
• Mike-Blog: Deinterlacing - Part 1 , Part 2
• Security systems: bad heredity - part 1 , part 2 , part 3
• Comparing deinterlacing filters by Michael Niedermayer (Michael Niedermayer).