Implementation of Support Vector Machine to Analysis Text Detection in Images and Videos

Abstract

Support Vector Machine (SVM) is a supervised learning model with associated learning algorithms that analyze and recognize data patterns used for regression and classification analysis. So many techniques have been used for the text extraction only from image or only from video. In this project we are using SVM technique for the analysis and extraction of text from both images and videos. Support Vector Machine has proved itself to be a good prediction technique than the Artificial Neural Network (ANN).SVM has been considered the fast and computationally less complex system than the ANN. This project tries to implement the SVM technique for the decision making part of the text recognition. The feature extracted from the video or image which tells that whether the text is present in the image or not. Text from videos is extracted by converting the video into frames. The main advantage of the present work is text extraction from image and video, effective in high dimensional spaces, can model complex, real-world problems such as text and image classifications. Compared to other techniques it will select the model size automatically. We are selected SVM because it has high dimensional input space, few irrelevant features, text categorization problems are linearly separable.