Handphone-based mobile image enhancement

Abstract

The paper reviews the lack of accuracy in existing document scanning applications while processing degraded document images and presents new enhancement algorithms for different degradations, including shadows, uneven illumination and out of focus blur. The proposed algorithm focuses on enhancing edges in text region and meanwhile suppressing the noise in background. Firstly, a pre-processing phase is introduced to restore the degradation to a certain extent. Secondly, adaptive unsharp masking is utilized to strengthen the text strokes. Lastly, thresholding and morphological operations takes place to binarize the image into black text and white background. The aim of this project is to build a novel scanner application that can boost the performance of document binarization using these enhancement algorithms. This application also possesses other relevant capabilities which includes boundary detection, cropping, exporting, sharing and text recognition.