Research And Application Of Data Completion Based On Truncated Kernel Norm

With the development of the Internet era in the 21 st century,data sharing and data acquisition become more and more convenient,and the amount of data we have is also increasing.So how to understand and analyze the data has become an important issue of concern to industry and academia.At present,although we can obtain a large amount of data relatively easily,data is sometimes unavailable.There are many factors leading to data loss,such as unreasonable collection methods and information loss in the process of collection,or some subjective and objective constraints.Since acquired data may cause missing or doping errors,to ease this problem,in this paper,we focus on data completion and recovery.In order to restore and complete the data,based on an overview of data completion theory,we propose a matrix completion algorithm based on truncated kernel norm of matrix.In this way,the original low rank problem can be approximated effectively.And we extend it to the following applications:1)Considering the application expansion of truncated kernel norm,we apply it to high dynamic range imaging and redesign the optimization algorithm.In this way,a more efficient truncated kernel norm algorithm is proposed,and the image restoration effect performs better than other methods.2)The truncated kernel norm of matrix is further extended to the data completion algorithm of tensor,where a truncated kernel norm of tensor is designed and optimized by using the alternating direction multiplier algorithm.The image restoration experiments show that the proposed algorithm has better recovery performance than the traditional algorithm.