Graphic-object detection in documents

Graphic objects generally occupy a good portion of any well written document. This is especially true for magazines articles and also for research documents. These objects also consume a lot of ink while printing, and would amount for large percentage of ink usage in general as compared to text printing. The goal of this paper is to smartly detect the regions of graphic objects and/or regions where lot of ink would be used while printing. These objects could then be selectively masked out while printing, and given as an option while printing is selected. Alternatively a low resolution of the same image can then be embedded into that region for completeness. The contribution of this paper is this novel idea of reduced `ink' printing and two simple approaches to robustly and efficiently detect the graphic objects. We have tested the proposed approaches on few sample test document pages from IEEE magazine. Our results are very promising. This technique and also could lead to Eco-friendly printing solutions if adopted.