Font Size: a A A

The Reserach Of Document Images Mosaic Taken By Multiple Cameras

Posted on:2013-09-13Degree:MasterType:Thesis
Country:ChinaCandidate:C ZhaoFull Text:PDF
GTID:2248330374988795Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
With the development of multimedia technology, the digitization of paper documents has become a trend, but for some large documents (such as newspapers, posters, etc.), a single image can not get higher resolution by the premise of guaranteeing the image quality. Hence, it’s necessary to capture document images many times, and use image mosaic technology to stitch them into a high-resolution document image. In this thesis, the problem of document image mosaic obtained by multiple fixed lens is studied, and the main works lie in deformation correction, splicing and fusion of document images by considering the characteristics of document image.Nowadays, deformation correction method of image does well in the distortion correction in the image, but the distortion between images is difficult to get corrected. Aiming at this problem, the thesis proposes an approach using the calibration standard drawings to calibrate the position relationship of lens, calculates the correction parameters, and corrects the distortion of document images. Firstly, strike the center of gravity of the lattice in the standard diagram, and generate the correction template. Then calculate the offset of center of gravity and the offset of the whole diagram by the bilinear interpolation method. Finally, achieve the correction of document images using the calibration parameters.The document which is rich in characters requires precise splicing, but the methods based on common template matching or features are prone to generate dislocation and ghost during the mosaic of document images. To solve the problem above, the method of document image stitching based on dynamic split line is put forward in this thesis. This method tries to find the best dividing line in the overlap region of image to achieve accurate splicing of document images.The light, color and other conditions of images can not be completely consistent, and accordingly illumination differences and stitching traces (seam) appear in the image which has been sutured. According to this point, the thesis proposes a light adjustment method based on the standard model to adjust the illumination of the document images. Firstly, design a model for light standards. Secondly, strike the light adjustment parameters from the standard sample image and the adjust template. Finally, the parameters are used to adjustment the illumination of document images, and then the weighted smoothing method is used to smooth the overlapping regions.The experimental results illustrate that the proposed method provide an effective way of deformation correction,splicing and fusion.
Keywords/Search Tags:mosaic of document images, deformation correction, best dividing line, image fusion
PDF Full Text Request
Related items