Unsupervised Enhancement Model for Brightness Equalization of Small Celestial Body

The economic and scientific value that small celestial bodies (SCBs) offer humanity are the main motivations for close exploration of these bodies. However, optical autonomous navigation becomes challenging due to the light variation caused by the rapid spin of SCBs. In this paper, we therefore propose a brightness equalization method, which consists of an Illumination Components Estimation Module (ICEM), a Brightness Self-Calibration Module (BSCM), and a Brightness Equalization Curve Optimization Module (BECOM). First, the adjustment factor and illumination components are estimated separately by BECOM and ICEM. These factors are input into the brightness equalization curve to remap the illumination components and further achieve brightness equalization of SCBs images. Next, we designed BSCM to calibrate the illumination components further. Considering the high coupling between the illumination and reflection components in SCBs images, we estimate the illumination components multiple times by the modules above. The input for each estimation is the output of the previous one, enabling weight sharing in a cascaded learning process, reducing the model parameters and accelerating network inference speed. Finally, we conducted extensive experiments on SCBs images, the results confirm the superiority and effectiveness of our proposed method.