동영상 자막 자동 배치 시 발화자 시선에 따른 위치 보정 알고리즘

Abstract

Conventional unedited video subtitles are placed in a certain fixed location. In preceding research, we developed an automatic video-subtitle placement system which places subtitle not in the certain conventional fixed location but near the speaker using our newly determined concept of face & head area. In this research, we devised correction algorithm of automatic video-subtitle placement which depends on the speaker’s gaze, and mathematical methods were used to correct the face & head area more elaborately. The developed automatic video-subtitle placement system using corrected algorithm was tested for placement of subtitles with several types of video genre. Comparing former and corrected algorithm, the performances showed overall 65.67% with subtitles’ coordinates change rate and overall 6.14% with subtitles’ final location change rate.