Research For Geometric Distortion Correction Technology Of Text Image

With the development of Internet technology and the popularization of Personal in every family, pictures and document data are transmited and dealed with in form of digital because digital information need less room for storage. Sanner as a digital tool has been used widely. During the course of scanning, many distortions can be caused, at the same time noises is inevitably. All these will put bad influences on image recognization and further process. we put an emphasis on geometric correction after preprocessing. So we have many works to dispose image before image is recognized .In this paper have two problems must be resolved, mainly involve nonlinear geometric distortion and linear geometric distortion .Nonlinear geometric distortion: Hough transform is used in this paper in order to detect slope angle. New method is adopted based on the property of Projection method can reflect black pixel distribution. some image block is segmented from the whole document image aim to improve the speed of Hough transform during detecting angle.Linear geometric distortion: some primary research is carried about this geometric correction method. At first analyze two correction method always used, at last combine their merits, A new method is introduced: wipe off the shadow of document image and correct the geometric distortion.