A new algorithm for text localizing using wavelet thresholding

Text data in video frames has contained important data for understanding. We with using: wavelet thresholding features, new projection profile method, and applying convenient filtering for colored JPG images, directly in compressed domain extract and localized horizontal text from complicated background. For this propose through various edge detecting methods, we offer detectors whose based on wavelet procedures and very different filters from soft to hard wavelet threshold. Our proposed method from one side aimed very higher accuracy than existing method and from other side is very fast and is low time consuming less method.