College of Sciences, Hebei United University, Tangshan, Hebei, China

---

Abstract—To study the leaves quality, this paper proposed two efficient models to analyze leaf quality, which 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 model, 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. 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.

Index Terms—factor analysis, snowflake theory, misjudgment probability, error evaluation

[PDF]

Cite: Lihui Zhou, Jiajia Sun, Juanjuan An, and Jun Long, " Research of Leaf Quality Based on Snowflake Theory," Journal of Computers vol. 8, no. 6, pp. 1520-1527, 2013.