The morphology features of leaf play an important role in plant taxonomy. A scheme was designed to acquire feature data of leaf shape automatically according to the requirements of feature extraction in angiosperm taxonomy. Firstly, a scanner was used to acquire the leaf image. Then the image preprocessing was conducted on the raw image, which included removing background, removing leafstalk and emending the position of lamina. Finally, feature extraction of leaf shape was performed on the lamina image to get the feature data of the Aspect Ratio and the Position Accuracy. Combining the feature data with the knowledge of plant taxonomy, the basic shape of the leaf could be identified. The test results showed the scheme was feasible.