Assessing The Document Image Quality Metrics for the Camera Captured Printed and Handwritten Documents

Abstract

Quality of an image determines whether it can be used for human analysis or any information retrieval process. Document Image Quality Assessment (DIQA) is one such method of quality assessment of documents, which consists of a complex process from data collection to model training g and obtaining the result. This Document Image Quality assessment (DIQA) has gained a prominent need in the human world today as it is used in the medical and education sector. This work aims to develop a model which can be used to access the quality of camera-captured document images (handwritten images) based on different handwriting styles which helps classify the document and predict the image quality. Overall document quality assessment based on the degradation angle of projection of camera based on textual features. The proposed model attains a machine-learning accuracy of 51.15%, and deep-learning model accuracy of 89.42%.