Image Reconstruction Of ID Card Based On L₁ Norm

Because our country’s ID card photo collection have been fixed standard, after process-ing after the photo ID card, because of compression and a series of transformations, lost most of the information, lead to reduce the quality of the image, image is very fuzzy, also because of the many requirements, restrictions to be collected, resulting in citizen’s photo ID card and I have a certain gap, but also for their life causing distress. The purpose of this paper is, through by the volunteers of the original identity certificates of the eye image and passport of the ocular image is composed of image sample database to learn a set of regression coef-ficients, and then reconstruct the high-frequency details and with living close to the photo of TD card.In order to solve the practical problems, this paper to volunteer ID photos and docu-ments photos based, respectively, structure the corresponding dictionary D1D1and D2 D2, and dictionary D1D1by optimizing model sparse representation to be reconstructed photo ID card f.The model is l1lnorm optimization model, namely to l1l1 norm instead of the difficult l0 10 norm as sparsity constraints, this paper through the orthogonal match-ing tracking algorithm (OMP)solving the optimization model and a set of sparse coefficient a. Combined with the set of coefficients a and dictionary d D2, close to the passport of the reconstructed pictures.In the experimental stage, this paper uses the above ideas, the most concentrated facial information as the object of reconstruction, through the above method, and finally get the reconstruction of the eye image. In the eye image and the original identification certificate of the merger, the edge part of the merger appears more serious edge effect, in order to make the image more perfect, this article will use the based on wavelet image fusion technology to eliminate the adverse effects caused by the edge effect. Finally, we reconstruct a picture of the eye clear ID card image.