This paper proposes a method for aerial ocean images segmentation based on multi-gauss characteristic space cover learning.Firstly
we analyze the distribution characters of sea background images and find that they are diversity in location
direction and geometrical morphology but clustering and can be covered by one or more spheroids.Then
we use the multi-gauss model to describe them and get the number of gauss components adaptively based on the maximum Bayesian posteriori probability and 3 criterion.Finally
we segment the aerial ocean images series according to their cover learning results.The experimental results show that this method can get the cover learning model accurately and effectively and segment the aerial ocean images with high precision and low error in less time.