Impact crater detection based on regional segmentation using Chang'E-1 CCD data

A regional segmentation based methodology aimed at automated recognition of impact craters on lunar surfaces is presented, and this method performs comparable result using Chang'E-1 CCD image. The initial and crucial phase focus on regional segmentation to generate the candidate region, we process the data though, firstly, nonlocal means(NL-means) filtering for image denoising when attaining more edge information, secondly, edge detection with canny operator, and third, morphological processing to extract the individual region with attribute information. This is followed by the phase of filtration and template matching procedure, in order to remove a certain amount of fictional candidates, and detect the center and estimate the dimensions of the targets. Statistics analysis of the regional attributes is undergo to verify the effect of this algorithm and this algorithm receives good performance at the experimental site of the mare.