Exact agricultural crops identification and planting area measure depend on not only classifiers but also training samples imported into classifiers.The effect of training samples

including sample quality and quantity

is greater than that of classifiers on measure accuracy.In order to probe into the effect of training sample on agricultural planting area measure

a representative wheat sowing section was taken as an example.A standard sample set was constructed based the popular TM image

SPOT-5 image with high resolution and field data from GPS.Then training samples with same features and different numbers were respectively input into six different classifiers to classify image and identify wheat

which are spectral angle mapper

parallelepiped

mahalanobis distance

minimum distance

maximum likelihood and support vector machine.Moreover

the experimentation was repeated ten times under different sample numbers.The results indicate: ①Ten measure results fluctuate from six methods under different training sample numbers.The fluctuation is reducing with the increasing of training sample numbers

which is greater in more fragmentized plot.There are some random errors in the current common used method of wheat planting area measure

which always uses one time classification result as the final value.In the task of agriculture planting area mensure

samples should be used as many as possible in classification to advance the result stability and the average of lots of classification results is advised to counteract the random errors.② The wheat pixel numbers（region whole area） differ each other among six methods under the same training numbers.And the diversity is larger in area with complex plant structure and fragmentized plots.Different classifier has different response to the same area

while same classifier has different response to different area.But it can be judge and verdict

which result is nearer to the true value

through analyzing the product accuracy and user accuracy of wheat.