Non line of sight imaging technology is a new 3D imaging technology for targets out of sight.The spatial and temporal distribution of the light intensity of a hidden object is collected by a detector mainly through the laser irradiation of the intermediate plane,and then the image is reconstructed by a reconstruction algorithm.Because it can only detect data at sparse (partial) angles,and cannot obtain all the information of hidden objects,which generates low precision in image reconstruction.In order to solve the problem of sparse angle data,a back projection modified maximum likelihood expected value maximization (BP MLEM) algorithm is proposed in this paper,which can effectively improve the accuracy of 3D objects reconstruction in non line of sight imaging.Compared with the filter back projection algorithm,the SSIM value of reconstructed result image is increased by 0.147、0.1607、0.22 and 0.2906 respectively after the 3rd,4th,9th and 20th iterations.Compared with MLEM algorithm,SSIM values are improved after several iterations.When the number of iterations continues to increase,the SSIM of the reconstructed result image obtained by MLEM algorithm tends to be stable.While the SSIM value of the proposed algorithm keeps increasing,indicating that the algorithm has better stability and is not easy to cause over convergence problem.