Objetive: In order to eliminate the cloud and mist from remote sensing images

a new non-local means algorithm is proposed to process the cloud and mist noise from remote sensing images. Method: First

based on the gradient feature under the shadow of cloud and mist in the remote sensing images

it finds that the intensity declines obviously while the gradient only has a little change

so it couples gradient information when compute the weight. Then

computing new weights uses redundant information in image sequences. Finally

restoring the image sequence using the new weight. Result: The proposed algorithm can restore remote sensing images without the motion estimation of the cloud and camera as well as the noise model. Two remote sensing image sequences were carried out by UltraCamD in Xinjiang and Shanxi area in China

and results show that relative to traditional algorithms

the quality of restored image improves significantly by this algorithm

and PSNR improves more than 9dB. Conclusion: It shows that this algorithm can restore images that coverd by thin cloud and mist effectively.