Novel Data Mining Techniques for Agricultural Data Sets

Abstract

Agriculture is critical to the development of the economies of developing countries. Food security will be aided by the high productivity of food yield. With global warming, however, achieving a high yield is difficult. Natural disasters such as drought, cyclones, and floods cause significant crop losses around the world, resulting in the death of farmers and people. A farmer needs funds for disaster management to be released quickly. As a farmer plants various crops in the same area, allocating the appropriate cash is difficult. As a result, crop identification methods that are efficient are required. For crop identification, hyperspectral imaging (HSI) is a useful technique. Crop recognition in the agricultural context has recently received a lot of attention. Despite this, agricultural identification using HSI has some limitations that must be addressed. First, the issue of incoming HSI data having a potentially large dimension size remains unresolved. Second, HSI data show that different crops have a lot of texture, form, and spectral signature in common. Finally, noise in HSI has a considerable impact on the accuracy of the current HSI crop categorization model. The HSI is made up of hundreds of Narrow Bands (NB), which are continuous and have a high spectral correlation. As a result, space and time complexity are introduced, resulting in considerable computing overhead and the Hughes phenomena while processing these images. newlineTo improve the effectiveness of hyperspectral image classification, dimensional reduction techniques like band-selection and feature-extraction are used. For lowering the size of features in hyperspectral images, existing approaches mostly used Independent-Component-Analysis (ICA) and Principal-Component-Analysis (PCA). The derived features remain independent with ICA-based hyperspectral agricultural classification approaches; however, ICA has a large computing overhead and does not ensure spatial information retention. When compared to ICA-based techniques, PCA-based hyperspectral agric