Hitachi's New Technology Recognizes Characters in Video Subtitles

E-Mail Article
Printer-Friendly
Tweet This
Digg This
Share this with friends on Facebook
Buzz Up!
Mar 18, 2008 14:57 Takuji Imai, Tech-On!

Central Research Laboratory of Hitachi Ltd developed a technology that recognizes characters in the subtitles of videos such as television shows.

The new technology enables to search characters in subtitles so that users can quickly find the scenes that they want to watch. There has been a technology that recognizes characters in subtitles. But it has been difficult to recognize characters when the colors or brightness of them differs in the same line and when they are mixed with frames or graphics. The new technology realizes a practical recognition rate even in these cases, Hitachi said.

The company has not decided when it will put the technology to practical use. The details of the technology will be announced at the 2008 IEICE (Institute of Electronics, Information and Communication Engineers) General Conference, which takes place from March 18 to 21, 2008.

The company conducted an experiment of applying the technology to about eight hours of videos including sports videos and recognizing people's names in the subtitles. As a result, a name or names were found in 329 subtitles, and names were correctly recognized in 93% of those subtitles.

In the experiment, Hitachi used videos encoded in MPEG-2 and used a PC to extract and recognize characters. A series of processes where characters in the subtitles of a video are recognized and compared with pre-registered names can be conducted at about 400-MHz clock cycle, the company said. A main storage capacity of about 20 Mbytes is required.

The characters in the subtitles are recognized in four steps. (1) Rectangular areas including subtitles are extracted from a video. (2) Character-string areas are extracted from each line in those areas. (3) The character-string areas are divided into character areas, and characters are extracted from each character area. (4) The character strings are compared with keywords that are prepared beforehand.

This time, Hitachi newly developed (2) and (3). In the second step, the company devised a system to remove, for example, graphics. In the third step, the company enabled to properly extract characters whose colors or brightness is different in the same line.

FPD Internatioan CHINA 2011/Beijing Summit
Microcontrollers
Analog