A System for Restoration and Structural Retrieval of Documents

Abstract

This paper describes a new technique for document retrieval using a web camera in an office environment. The architecture of the aforementioned method consists of three main stages: segmentation, restoration and retrieval. Firstly, the document image is taken with a web camera and the segmentation process is applied to locate the four corners of the document and isolate it. Then we proceed to document restoration by applying filtering, skew and curvature correction as well as removal of any redundant object that does not belong to the document. In the third and final stage, a feature vector is extracted and is compared with the documents of the database.