Research On Correction Algorithm Of Document Image

Book document refers to the binding information,bibliography and other printed material.One image scanning through the artificial page and are collected by the camera.In one image scanning process,due to the inclination of document image different document placement may be caused by.On the other hand,due to the document bound volumes and finger pressing reason document pages bent,and at the borders of the image will finger region exists.Finally,the document curved surface light re:flection angle caused by different but also may cause uneven degree of image contrast.In order to solve the above four problems,this thesis deeply studies the algorithm of image correction.For volumes document image tilt,according to the documents about the boundary information of rotation correction;for image plane bending,according to the document of the upper and lower boundary curve,to establish the mapping relationship between flattened image coordinates of points and,application data interpolation method computation flattening after pixel values.For existing in artificial finger image,the connected components labeling to identify;application of image threshold segmentation method to solve the image contrast is not obvious problems.The main research work includes the following three parts.Firstly,the image rotation correction algorithm based on Radon transform is proposed in this thesis..First,using Prewitt edge detection operator detection volumes image boundary line,according to the image edge features,select points on the circle to the left of the image to image rotation correction reference point;then,the images do Radon transform processing,take accumulation and maximum value inclination angle.Finally,based on the inclination angle of image rotation correction.Secondly,according to the book image curved surface characteristics,this thesis presents an image smoothing algorithm based on space coordinate transformation.First,calculates the volumes of the left and right image boundary,according to the coordinates of the left and right boundary line to calculate the image of four boundary points;secondly,it is estimated volumes of image boundary fingers may be occluded volumes image boundary,need to estimate the occlusion of the lower boundary;then,to establish the space coordinate transformation.Finally,using the bilinear interpolation calculated value mapping coordinates.Thirdly,this thesis studies the problem of information in the image and the image surface of artificial finger into the problem of uneven contrast.In order to solve the above problems,first of all,the image of median smoothing,the smoothed image and the original image subtraction from the absolute value,the subtraction of gray image;secondly,fixed threshold segmentation of gray image;then,the segmented binary image connected body marker,find where the finger position according to the features of finger;last adaptive threshold segmentation for gray image and finger region is set to zero,the output image.Through experiment and Simulation of this algorithm.The experimental results show that the effect is very good,volumes scanned image of linear distortion,nonlinear distortion,finger area and brightness uneven problem solved.