INTELLIGENT PRINTER

Abstract

Paper has traditionally been a primary medium for transmitting information. Compared to electronic media, reading and interpreting information on paper is easier and simpler to find mistakes. Despite the increasing digitalization of information, there has been a rise in paper usage compared to previous times. Our reliance on paper extends to printing various materials such as agendas, meeting minutes, spreadsheets, retail catalogs, brochures, bills, receipts, notices, drafts, and final reports. In spite of the efforts for a paperless society and the rise of electronic media, paper production has seen minimal change. Mostly due to the presence of confidential information often printed on office paper, which needs secure disposal. Even If the paper is disposed of and remains in a landfill, it imposes potential harm on the environment. The paper industry is in fourth place among the largest emitters of greenhouse gases. It may also lead to issues like deforestation, soil erosion, losses in biodiversity, pollution of water resources, and consumption of energy. The conventional paper recycling requires continuous small batches of virgin wood fiber because recycled pulp cannot be used more than 4 to 5 times. We are introducing a new recycling technique to replace conventional methods, which would eliminate multiple stages in paper recycling, leading to a significant reduction in resource utilization. Our research uses the obscuring technique to conceal the printed content behind a material that matches the color of the paper. It is important to comprehend and analyze the document's structure to conceal it adequately. We applied classical image processing techniques, binary segmentation, and deep neural network layoutLMv3 model to segment the content of the document image. We prepared the mask of the document's content using the layout analysis and printed it again on document paper, the paper that required recycling. LayoutLmv3 is trained on the PubLayNet dataset and evaluated on the custom dataset and ICDAR SmartDoc mobile images. Our recycling method provides instant and multiple recycling facilities, which cut off the high energy and resource demands, with one time recycling saving 50% and second-time saving 75% of resources.