Abstract: In order to solve the quality problem of remote sensing image degradation which is caused by fuzzy and noise in the process of push-broom full-color TDICCD space camera acquisition, transmission, and storage image, a new remote sensing image restoration method based on modulation transfer function (MTF) was proposed. The method in this paper restored point-spread function PSF by MTF, and took it as degradation model for image restoration. The ability of noise resisting of MTF extraction was raised in this paper via enhancing the local image by homomorphic filter algorithm before MTF extraction. Then by using the two times restoration method, MTF error which is caused by adding enhancement filter directly was avoided. Finally, the effect of remote sensing image restoration was evaluated by mean, contrast and edge strength. In experiments, the recovered image grayscale average increases 45%, average gradient is 3.5 times and the edge strength is 1.75 times greater than the values before restoration. And the area which is contained by MTF curves increases. The experimental results show that all above evaluation indexes are better than the original image. In conclusion, the image restoration algorithm of space TDICCD camera has certain reference significance in improving image quality.