This paper proposes a new approach to estimating micro-debris polulation from solar array images. As for Geosynchronous Earth Orbit, space debris population has not been sufficiently evaluated because of long distance from the earth to the orbit. On the other hand, solar panels of geosynchronous satellites usually have so large area that a lot of micro-debris and cosmic dust can give some damages, which can be also effectively utilized as a dust counter. Aiming at in-situ debris observation by impact craters in the solar array images, firstly, a recovering method of real planar shape of the craters from slant view images is proposed. Secondly, to estimate micro-debris size distribution in the surrounding neighborhood in the orbit, a Bayesian approach based on statistical model which represents relation between the impact crater's geometry and the original debris' physical property is applied. This paper also demonstrates the proposed scheme through computer simulation and also discusses current issues to be solved.