This article presents methods of image analysis based on supervised learning and an algorithm consisting of two stages of determining the optimal classifier using a cluster ensemble.  At the first stage, the averaged co-association matrix is calculated using a cluster ensemble. In the clustering ensemble, we used a scheme of a single clustering algorithm that constructs base partitions with parameters taken at random. At the second stage, the optimal classifier is determined using the resulting kernel matrix as input data. Numerical experiments were carried out with real hyperspectral images. The experimental results showed that the proposed algorithm has classification accuracy comparable to some modern methods.