Study on Super-Resolution of Images Obtained by Micro Satellite with CMOS Sensor
Jun Xu, Sun Tingting · 2019
This paper proposes a new image reconstruction algorithm to improve the resolution of images obtained by micro satellite with CMOS sensor. Satellite with CMOS Sensor can obtain images of same ground object in consecutive integration times. Because of the low attitude stability of micro satellite, there are sub-pixel displacements for same target between these images. In order to measure the displacements, three image registration algorithms, which are Maximum Mutual Information, SURF (Speeded Up Robust Features) and Keren, are compared through simulation experiment. Then, an image super-resolution reconstruction algorithm named Weighted Average Interpolation is proposed. Based on two sequential low-resolution images, one high-resolution image is acquired through Weighed Average Interpolation. Simulation result proved that PSNR in synthesized image is enhanced by 1. 44dB compared with that in bilinear interpolated image. Test chart imaging experiment showed that Average Gradient, Edge Intensity and Space Frequency of synthesized image are all obviously greater than of bilinear interpolated image. The resolution of synthesized image is effectively improved by Weighed Average Interpolation.