Font Size: a A A

Study On Total Variation Regularization Models Based Color Filter Array Image Demosaicking

Posted on:2020-06-13Degree:MasterType:Thesis
Country:ChinaCandidate:M L LiuFull Text:PDF
GTID:2428330590956564Subject:Mathematics
Abstract/Summary:PDF Full Text Request
If we want to obtain all the color information with a digital camera,we need to go through the optical system and project the light onto three different sensor elements.By this way,the generated color image does not have mosaic phenomenon,but the alignment of three sensor elements is more demanding for industry,this digital camera is too large to easily carry and very expensive,so the common digital camera we adopt is single sensor.However,only one third color components is sampled at each pixel point by using this camera.The sample datas can be used to recover the full color image informations and the reconstruction process is called color image demosaicing.The traditional total variational regularization method adopts the second-order partial differential equation model,which effectively preserves the edge of the image,but does not take into account the correlation of the three color components,and it is easy to produce the staircase effect.To improve the image restoration quality,an improved total variational regularization model basing traditional total variational regularization method is proposed,which takes into account the the relationship of three channels,and introduces the grayscale term of color image as the second term of regularization term.In view of the fact that the traditional model is easy to produce the staircase phenomenon,a high-order regularization model is proposed,which replaces the second-order partial differential equation model with the fourth-order partial differential equation model,and effectively eliminates the step effect produced by the traditional model.The traditional regularization model can effectively preserve the edge of the image,and the higher-order regularization model can eliminate the staircasing phenomenon.Thus,a hybrid higher-order regularization model is proposed,which can effectively preserve the edge of the image and eliminate the staircase effect simultaneousely,then obtain a high-quality demosaicing image.In order to prove the availability of the model,the corresponding Matlabprogram was written.Conducting the numerical experiment by using color images,which verified the validity of the model and algorithm.
Keywords/Search Tags:Grayscale images, Staircasing phenomenon, The primal-dual fixed point algorithm, Demosaicking
PDF Full Text Request
Related items