Automated classification of soil images using chaotic Henry’s gas solubility optimization: Smart agricultural system

Abstract The advancements in automation and image processing techniques lead to the agricultural research a prime focus for the researchers. Soil quality prediction is one of the important factors for crops and it is tested manually by the farmers. Therefore, automated methods for soil prediction are very helpful in increasing the efficiency of this process. In this paper, a new optimal bag-of-features based automated soil prediction method is used to categorize the soil images into a set of predefined categories. In this method, an enhanced Henry gas solubility optimization(HGSO) is used to generate optimal visual words. To improve exploration behavior of HGSO, chaotic map based Henry constant is introduced. The proposed variant of HGSO performs effectively on considered benchmark functions with better convergence precision. Moreover, the proposed classification method has been tested and validated on soil image dataset having seven classes and experimental results show that the proposed method returns 80.58% accuracy which is better than other meta-heuristic based classification methods.