How to improve the reconstructed image quality using inherent prior knowledge of natural image is still a crucial issue in compressive imaging.In this paper

an efficient compressive imaging algorithm is proposed

which combines the sparse property of the entire image patches and the manifold property.In the algorithm

image patches are represented sparsely by a dictionary and approximated simultaneously by a set of neighbor points.These neighbor points are embedded in the same manifold with the original signal

ensuring that the recovered signal using sparse approximation is close with its manifold.In addition

the proposed compressive imaging algorithm which makes full use of the sparse properties of patches in any position of the image can reconstruct the high quality image in low sampling rate.