Abstract: In order to retrieve knowledge from the old and voluminous text books its necessary to develop a method to reconstruct the scanned pages texts while defection of these old text pages is very usual which causes two major problems, the shadow of text which is caused by the scattering of ink over the wet paper and the deformation of old paper which naturally occurs as a result of temperature and humidity variations and causes the text to be seen curved and unrecognizable. It’s the objective of this present work to represent a method resolving the two inconvenient. First, the border between the text and its shade should be determined utilizing a connected component analysis based noise reduction method. Second, the discovered borders are compared with the previously standardized letters edges and then the current text is identifiable. The method is experienced at several pages and results are displayed.