The Algorithm Research Of Image Restoration Based On Total Variation

Image is an important source of information.In the process of image collection,processing,storage and transmission,image quality is degraded due to the imperfections of imaging system,storage equipment,transmission medium and processing method,this phenomenon is called image degradation.It has a great problem in some aspects such as the image of observation,analysis and extraction of the details.Image restoration is the recovery of the original image from the degraded image by some certain methods,means and rules.Therefore,it has a very important significance for the image restoration problem in the field of image processing.As the Total Variation(TV)image restoration model can be very good to maintain the edge of the image and remove the noise and other unnecessary details,which is widely used in image restoration problems.The TV image restoration model generally consists of two parts,i.e.regularization term and fidelity term.But this model is easy to produce stair effect.Therefore this thesis proposes a new image restoration model after modifying the regularization term.In this thesis,we use the grayscale image as the research object,and use the alternating direction method of multipliers and the primal-dual algorithm to optimize the new model and improve the image quality.The work of this thesis is:(1)The research background of the image restoration problem is introduced in the first section and the researches are mentioned in recent years.There are some basic knowledge in image restoration,then introduce several types of image blur and noise type and give the image restoration quality evaluation standard.Finally,the model of this paper is introduced by the total variation model.(2)For the alternating direction method of multipliers and the primal-dual algorithm,the basic principle is given respectively,then apply these two algorithms to the new model.(3)There are some experimental results of the alternating direction method of multipliers and primal-dual algorithm,then the two algorithms are compared with other methods,finally compare the alternating direction method of multipliers and primal-dual algorithm.The experimental results show that the primal-dual algorithm is superior to other methods and the alternating direction method of multipliers.