Weather Classification Based on Hybrid Cloud Image Using Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA)

Abstract

Changes in weather and climate conditions have consequences on various sectors of life and greatly affect the activities of human life. Therefore we need a system that can detect weather conditions based on cloud imagery. Finding methods to detect weather conditions at one time with image processing is a new innovation that appears in current weather modeling. This is driven by the high need of various parties to conduct research in detecting a condition carefully and without having to observe it directly. In this study a climate condition classification system will be designed based on cloud imagery using the Hybrid method, namely PCA + LDA. All cloud imagery will be grayscale then feature extraction and cloud classification process using Euclidean Distance. Based on the tests carried out, the system produces an accuracy rate of 96%. The predicted weather conditions are bright, cloudy, and rainy conditions.