Abstract:

A method for autonomous detection of crater from gray images is proposed. The original image blocks are census-transformed. The obtained histograms are used as feature vectors. Principal component analysis is performed to compress the feature space. Pattern classification is done with support vector machine. Craters of different sizes are detected by reducing and magnifying the crater candidate area. The results show that the proposed method can effectively detect craters of size greater than 20*20.

Key words: crater, Census transform, principal component analysis (PCA), support vector machine (SVM)