Text Extraction for Android App

Abstract

Inspired by a popular technology of text recognition. This is an Android based application that developed for removing non-text objects, background image, recognize and extract the text. This application recognizes the text from the captured image by using the image convolution, down-sampling, filtering, averaging and histogram threshold. In this project, 9 text filters are used in the convolution calculation to raise the accuracy for text segmentation. No matter the text size, front and character styles, most of the text can be extracted from the photo that consists of some complex non-text objects. With the help of the text recognize application, users can easily extract the text from any image format. For example, recognize the dish name from the menu and the personal information from the business card, which makes life much easier.