Planetary Image Live Stacking via Phase Correlation

During planetary imaging, the complicated imaging environment, atmospheric turbulence and the equatorial mount's inaccurate compensation for the Earth's rotation will cause image degradations including noise and deformation. A low-cost solution is image stacking which stacks low exposure frames to achieve long exposure result or averages long exposure frames to reduce noise level. The most critical part of live stacking is image registration, which could effectively correct deformations among frames. There are already some offline stacking algorithms and products. However, in order to obtain better astrophotography experience, live stacking is a better choice. Astronomical live stacking emerged in deep-sky imaging fields where feature-based image registration method is preferred. However, in planetary image registration, feature-based methods may fail because of that features cannot be extracted stably and repeatedly among different images. This mainly is because of that planetary images lack enough texture features to be extracted. Therefore, area-based registration is the better choice. In this work, we proposed an algorithm based on phase correlation to implement planetary image live stacking. Benefiting from fast Fourier transformation, phase correlation has received widely applications in remote sensing, biological and medical imaging, etc. However, with the increase of image resolution, the computation cost increases too. That limits its application in real-time processing tasks. In this work, we utilized the phenomenon that the signal power of down-sampled images always concentrates in a few coherent peaks and extended the original closed-form solution to implement high resolution image live stacking with subpixel accuracy. Experiments on planetary video stream illustrates that the proposed algorithm has satisfactory results with lower computation cost compared with the original phase correlation.