Now a days digital video is very popular as an exchange mediumdue to large improvement in video recording and compression techniques andincreasing of network-speed. Therefore audiovisual recordings are used more frequentlyin e-learning and e-lecturing systems.
OCR from videography is a technique thatcan locate any text inside a digital video file via reading and automatic extraction of any notes and captionsthat gives the actual information (like – the names of people, places ordescription of objects etc.) about the video being presented. Detecting the video-contentrequires many technologies: image detection, language processing, searchstrategies, video segmentation and filtration etc. Reading the extracted notesand captions gives more appropriate information to understand the video-content.Applying OCR on video and combining the results with various detecting techniquescan improve the detection result. Although integrated character recognition intext-based videos is greatly needed. Automatic character segmentation was performedfor titles and credits in motion picture videos in however;papers have insufficient consideration of characterrecognition. There are similar research fields which concerncharacter recognition of videos.
In character extraction fromthe car license plate using video images is presented andcharacters in scene images are segmented and recognizedbased adaptive thresholding. While these results are related,character recognition for the video presents its owndifficulties because of different conditions of title charactersize and complex backgrounds. In video caption resolutionof character is lower; also, the background complexity ismore severe than in other research. The first problem is lowresolution of the characters. The size of an image is limitedby title number of scan lines defined in the NTSC standarda character of the video caption are small to avoidocclusionof interesting objects such as people’s faces. Therefore, theresolution of characters in the video caption is insufficientto implement stable and robust Video OCR systems.Another problem is the existence of complex backgrounds.
Characters superimposed on videos often have hue andbrightness similar to the background, making extractionextremely difficult. These problems in video OCR haveopened an area for research.Video OCR is a technique that can greatly help to locatetopics of interest in a large digital video via the automaticextraction and reading of captions and annotations. VideoOCR process and all the process modules required in videoOCR are explained in section III. Applications of videoOCR are explained in section IV. Conclusion based onrelative work is explained in chapter V.