Relationship between paddy soil moisture and image texture

Different soil moisture content gives different soil texture. Texture of dry soil is rougher compared to the moist soil. In this paper, statistical analysis has been done to the paddy soil images to determine the relationship between soil moisture content and its texture. Two hundred and sixty five paddy soils samples with the soil moisture content range from 12% to 80% were used. Eight texture features namely mean, variance, homogeneity, dissimilarity, entropy, contrast, second moment and correlation were extracted from CIELuv soil image. CIELuv was used as it has the advantage as a device independent where the soil color is less affected by its surrounding. Results from the ANOVA test has shown that only mean did not give significant relationship. Entropy and second moment give the value of correlation greater than 0.5. Second moment give better performance in predicting the soil moisture content using linear regression model with the value of R2 = 0.4