Crop monitoring system using image processing and environmental data

Abstract

We propose advanced crop monitoring system using image processing from environmental data for faster and better yield of agriculture. In the system, we have used colour segmentation and binary masking to analyse and differentiate between various stages of crop plantation (unripe, ripe and diseased) via aerial images. To further improve accuracy, we have used different sensors to detect important factors like the temperature and moisture of the air and soil. This is going to project precise areas of the affected regions and can be taken care of immediately after detection. Furthermore, using Support Vector Machine (SVM), we have been able to classify different types of leaf diseases up to 98% accuracy. We have also used other algorithms, like K-mean cluster and L*a*b* colour space, to detect the diseased part of the leaves. This will ensure the farmers to provide the right type of treatment for the plants. We have taken this initiative in the perspective of Bangladesh where a huge amount of population requires huge amount of food supplement and hence we need a faster and safer monitoring system to meet the agricultural needs of the nation.