부산대학교 도서관

Soil moisture is important parameter in agricultural and hydrological fields. A measure of unconfined compressive strength (UCS) and moisture content in soil decides soil strength. This analysis is done in civil engineering laboratory which consume more time and man work. This paper explained about survey in determination of UCS and moisture content of soil by image analysis. The soil sample image is taken in laboratory and analysis is made using various algorithms. The various methodologies are followed to detect the soil moisture. In image acquisition satellite images, un-manned aerial vehicle, hyper spectral images, thermal images, digital images are used for soil moisture prediction. Various segmentation methods are used to extract the region of interest. The colour models as RGB, CMY, HSI are extracted at different moisture condition and texture feature as GLCM, haralick are measured to determine the correlation between intensity of the pixel. The colour and texture formation differs when moisture condition of the soil changes. The measured features are reduced and classified using artificial Neural Networks (NN), SVM, Random forest and various classification methodologies. The classification method differentiates the soil at different moisture level condition. The different techniques to measure soil moisture and importance of soil moisture is discussed in this work.