The Shape Classification of Snowflakes Based on BP Neural Network

Abstract

In view of the problem of snowflake shape classification in the basic research of meteorology, a kind of snow flower shape classification method based on BP neural network is proposed in this paper. First we preprocess the snowflake images and extract the contour characteristics of the snowflake; on the basis of the contour characteristics of the snowflake, we can obtain six morphological parameters such as the axis ratio, rectangularity, convexity area, convexity perimeter, form parameter and density of snowflakes, and the parameters will be divided into three kinds of snowflake shape categories; finally we analyze the nonlinear relationship between the shape features and classification results, to design the BP neural network classifier that can divide the snowflake images into 3 types. Experiments show that the recognition rate of this classifier can reach 91.67%, which can provide reliable data support for subsequent research on the relationship between snowflake physical structure and artificial intervention in snowfall.