Abstract: This paper presents one cloud detection method over ice-snow covered area in Antarctica.On account of different texture features of cloud and ice-snow area,five texture features are extracted based on GLCM.Then non-liner SVM is used to obtain the best model from training data.The experimental result shows that this algorithm performs well in cloud detection in Antarctica,especially to the thin cirrus.Furthermore,when images are resampled to the quarter or 1/16 size of the full size,cloud percentage does not change greatly,while the processing time decreases exponentially.