Preliminary simulation of robot on script detection from camera images

Abstract

Here is a lot of information is available in a photo images such as advertisement, book cover, banners and many more. Images with text are widely available because of efficiency and low cost of digital portable devices, which provide chances to manual transfers of a document images. Analysis techniques of manually transferred images could serves as a reference and starting point of further technique development but the method cannot be used directly on images captured using cameras. Camera pictures can be problematic with blurry, low resolution, distorted, disorientated images, apart from, complex interaction between content and background. Therefore, Optical Character Recognition (OCR) technique was used to change printed text into editable text that is convenient and accessible in various applications. However, OCR accuracy depends on text pre-processing and segmentation algorithm. Hence, this manuscript to introduce OCR Tesseract method and the history of OCR Open Source Tesseract system, its architecture and outcome of trial on various type of images to determine efficiency of OCR Tesseract and accuracy ratio of extracted images from camera.