Compressed imaging avoids pursuing large array devices in infrared band, and it can solve problems like the heterogeneity which is difficult to eliminate during the image acquisition, low sound to noise ratio, higher and higher cost of acquiring, transmitting and storing images in the aerospace imaging applications and so on. The principle model of the compressed imaging system is analyzed in details, an imaging principle prototype is built, and the experiments of image reconstruction are conducted by the gradient projection algorithm. The signal subspace analysis is introduced in quality assessment of the reconstructed image to estimate the signal to noise ratio of the reconstructed image. Experimental results show that the proposed estimation method is more accurate and effective.