Distinguishing between shape and text strokes is a critical task in recognizing sketch documents that contain both shapes and texts. As the stroke constitutional complexity of shapes and texts is different, we proposed a shape and text distinguishing method based on the calculation of the strokes’ entropy value. During experiments, we found the entropy of a document may vary with changes in stroke size, and we implemented an adaptive resampling method to handle these problems. Our method produced a correct classification rate of 93.33% on the test data.