Puttaswamy have proposed the system that performs area based filtering to eliminate noise blobs present in the image. This method can detects only static superimposed text. It performs content based video indexing. This method is only applicable to MPEG videos.īaseem Bouaziz, Tarek Zlitni, Walid Mahdi explained automatic video text extraction. It reduces spatial and temporal data redundancies. Lifang Gu explained text detection in MPEG (Moving Picture Experts Group) video frames. The prerequisite for this method is that, text should have more contrast compare to its background. As text enhancement is not been used the extracted text can be noisy.Īrvind, Mohamed Rafi have explained text extraction using connected component based method. Text detection is performed using sobel operator and thresholding. Priti Rege, Chanchal Chandrakar has explained text image separation in document images using boundary/perimeter. This method fails when two different images having exactly same color histogram values. It involves checking color histogram for each frame against the histogram of the next consecutive frame. Pitas have proposed the system that uses entropy based metrics. This method is sensitive to skew and text orientation.Ĭennekove, C. Priya have proposed combined edge based method.
Mati Pietikainem, Oleg Okun have proposed combined edge based text detection that minimizes degradation in extracted text and can work with images having complex background.Ĭ. The detected text from each frame is stored in text file.ĭatong Chen, Jean-Marc Odobez have proposed the system that minimizes character error rates and also removes noise from the character that greatly disturb the optical character recognition.
VIDEO TEXT RECOGNITION SOFTWARE SERIES
Proposed system converts video into series of frames and applies text detection and extraction on each frame. This video is provided as input to the proposed system. User downloads the video form the YouTube or any other website from which he wants to extract text. The working of the proposed system is very simple. Also the information in text can be edited if it changes in future or if user wants to add any additional information in it which is not possible in case of video. The main advantage of text file format is that it requires very small size as compared to the size of a video. In such case the proposed system helps user to get access to the information by converting to text in video to editable form. Once the user has watched the educational video, next time he may not want to go through the entire video as he has already watched it and reading the main points may be sufficient for him to revise the topic from that video. If the text from the videos is converted to editable form, it can be stored efficiently and it will be easier to access it next time. These videos contain text which adds information to videos and makes it more meaningful. YouTube is used widely used for news and educational videos. The proposed system will convert the text in the video into editable form which is stored in a text file. As the focus is shifting towards YouTube the paper has proposed system which makes it easy for user to access information contained by the text in these video in efficient and quicker way. Television has programs that are shown at a particular fixed time which creates time constrain for users. The main advantage of YouTube over television is that YouTube provides shows at users preference irrespective of time. With the rapid advancement in technology and the increasing speed of internet, the focus of the people is shifting from Television to YouTube. Keywords Frames, Text Recognition, MESR, OCR, Gray Scale, MPEG. The system will process the video and generate the text output in editable text file. The user will have to provide the video as input from which he wants to extract text. The main focus of the proposed system is on educational and news video. The paper describes the technique that aims at extraction of the text which occurs is video. If this text is converted to an editable form it becomes simpler and efficient to store useful information. But this information is not in editable form. The text in the video contains huge amount of information and data. Mahendra Patil Head of Computer Department Atharva College of EngineeringĪbstract Videos have become a great source of information.
Student of Computer branch Atharva College of Engineering
Text Recognition and Extraction from Videoī.E.
0 kommentar(er)
