Text Region Extraction from Punjabi News Videos using Feature Analysis

This paper presents a text region extraction approach based on feature analysis. This scheme consists of 5 main phases that are Frame Extraction, Edge Detection and Binarization, Fusion, Normalization and Feature Extraction. For Frame extraction MATLAB simulator is used where video reader is used to extract frames from the video according to the length of video. Edge detection and Binarization is processed on that extracted frames then both the detected edges and binarized image will be fused using PCA based fusion method. Fused image is then normalize and feature extraction is perform by using SIFT and SURF based algorithm. The results will show that extracted text region from the frames of the news videos.