摘要：To study the leaves quality, there were three parts in this paper. The first part was to classify leaves based on different shapes, Leaf shapes were classified from the macro and micro perspectives respectively. In the two perspectives, influential factors were extracted and analyzed by factor analysis and K-means clustering. After comparing clustering result with actual classification result, misjudgment probability is found to be very low. In the second part, snowflake model theory was proposed. The theory is high similarity between snow structure and tree structure, and the formation of the branch copies the exterior characteristics of the backbone. Then the growth process of a tree was simulated, after calculating the number of smallest branches through programming, the total number of leaves could be calculated out. In the third part, to estimate the tree leaf weight, two steps were divided. First step was to estimate the number of leaves using the snow theory. Second step was to estimate the area of single leaf. Finally, the area measurement model to flat leaf was set up to measure the area of the curly leaf, which was dividing the whole curly leaf into small pieces.
