The SEaTH method （ SEparability and THresholds） can select features and compute thresholds automatically based on Jeffries-Matusita distance to separate classes

which may cause information redundancy and affect classification results. In this paper

a new method based on SEaTH is proposed

and then the method that considers the Jeffries-Matusita distance

the inner class distance

and the correlation of different features is used in the classification of multi-resolution remote sensing images of Zhaoqing

China. The result indicates that our method can select more effective information than the original SEaTH algorithm

with accuracy in extracting farmland improved by 12. 26%

and total accuracy improved from 80% to 85. 26% . Such an improvement is significant to the classification when multi-temporal data are difficult to be obtained.