Spaceborne cameras are often called as detector resolution limited system. This is because the detector array generally suffer under sampling. Therefore, high frequency information beyond the detector sampling frequency will leak into the detector array, i.e. every remote sensing image will include some high frequency components. Since the platform keep drifting and vibrating even in the geostationary orbit, every remote sensing image contains unique high frequency information. By collecting those high frequency components from a sequence of images, a high resolution image can be reconstructed. This is the theoretical basis of super resolution technique. Moreover, with the widespread use of spaceborne CMOS array detectors, it is possible to obtain hyper-temporal data, which brings opportunities for spaceborne CMOS Cameras. A new hyper-temporal imaging mode for spaceborne CMOS cameras was proposed in the paper. By using a CMOS camera to continuously and quickly capture sequence of data, many frames of images within the same area can be extracted. By solving an ill-conditioned equation, high resolution images with improved quality can be achieved, that is, digital time delay integration TDI (Time Delay Integration), Modulation Transfer Function (MTF) and Super Resolution can be implemented at the same time. In general, engineers would like to set long expose time for spaceborne cameras to ensure SNR for remote sensing images. However, long expose time inevitable bring blur which severely decrease the quality of remote sensing images. The advantage of this new imaging method is that it can freeze the images to avoid blur as speckle imaging technique widely used in astronomy community. To reconstruct an improved quality and high resolution image, we need a good understanding of the whole process of capturing LR images. Since spaceborne cameras can only capture the reflected light from the surface of the earth and the reflected light suffers from the air turbulence and diffusion from the optical lens system. Therefore, mathematically modeling the image degenerating procedure is very important. As we all know that image restoration is an ill-conditioned problem. In terms of solving the ill-conditioned problem, a mixed sparse representations is used. In general, it is very difficult to find a common sparse representation for remote sensing images because of complicated ground features. In the paper, a remote sensing image is regarded as a combination of sub-image of smooth, edges and point components, respectively. Since each domain transformation method is only capable of representing a particular kind of ground objects or textures, a group of domain transformations are used to sparsely represent each sub-images. By using the generalized sparse representation, image restoration can be solved through the traditional L1 norm based optimal algorithm method the iterative thresholding algorithm. Experimental results based on the low-orbit optical remote sensing satellite OVS-1A, Jilin-1 video 03 satellite and the geostationary optical satellite GF-4 show that both the signal-to-noise ratio, image clarity and spatial resolution have been significantly improved. The proposed method holds promise to bring new remote sensing imagery products with high resolution of improved quality for satellites in orbit. Moreover, the method can also save the cost for future planned satellites by reducing the volume and weight of the optical camera payload.

hyper-temporal data;small satellite;Time Delay Integration(TDI);Modulation Transfer Function(MFT)