Research Of Super Resolution Of Depth Image Based On Sparse Representation Algorithm

In order to improve the resolution and quality of images,image super-resolution reconstruction technology emerges as the times require.This technology can not only meet people's visual needs,but also facilitate the follow-up application of images in many engineering fields.In recent years,with the development of high-tech such as machine learning and artificial intelligence,learning-based methods have been widely used in the field of image super-resolution reconstruction.The image super-resolution reconstruction technology based on sparse representation theory relies on its complete theory,which makes the reconstructed image obtain good results and further promotes the image super-resolution reconstruction technology in various fields.Development has played an important role.At the same time,in recent years,according to the characteristics of different kinds of images,scholars have combined reconstruction-based methods with learning-based methods,and achieved good results.The birth of such methods provides many new ideas for super-resolution reconstruction technology.Depth image is a kind of two-dimensional image which can reflect the structure information of three-dimensional scene.It connects two-dimensional data and three-dimensional scene.It has a great application in the current multimedia field,but its image quality is affected by many factors and can not be applied.This paper mainly solves the problem of low image quality of depth image.The main work and innovations of this paper are as follows:Firstly,a super-resolution reconstruction scheme of depth image based on joint sparse representation is proposed.Considering the characteristics of the depth image itself,this scheme constrains the whole reconstructed image as a priori information,and then proposes a multi-dimensional sparse optimization strategy.At the reconstruction end,the steepest descent method is used to solve the convex optimization problem,and finally the depth image is reconstructed.This method can constrain the structure of depth image well.The experimental results show that the scheme is effective.Secondly,considering that the depth image is finally applied to the restoration of three-dimensional scene,the information related to the depth image in the color image of scene is added to the prior information,and the multi-regularization strategy is adopted to establish the relationship between the low-resolution depth image and the color image of scene,and the relationship is applied to the reconstruction process to complete the reconstruction of the high-resolution depth image,in order to further optimize it.Finally,the global regularization constraint is introduced to remove the artifacts effectively.The experimental results show that the scheme is effective.