Segmentation of nutrient deficiency in incomplete crop images using intuitionistic fuzzy C-means clustering algorithm

Abstract

This paper introduces a new segmentation technique to segment incomplete nutrient-deficient crop images by imputing missing pixels. Usually, each image contains pixels holding information about intensity, but sometimes image can miss some pixels (that is, the pixel without an appropriate intensity value). An image with missing pixels is called an incomplete image. Intuitionistic fuzzy clustering algorithm is a useful tool for clustering images, but it is not directly applicable for incomplete images. For example, segmentation of nutrient deficiency portions in the presence of missing pixels leads to error in segmentation. Crop images with nutrient deficiency might have missing pixels due to inherent defects in imaging equipment or due to environmental conditions. In this paper, nutrient deficiency in crop images is segmented after imputation of missing pixels based on intuitionistic fuzzy C-means color clustering algorithm. Experiments are performed on various incomplete crop images. Through the derived results and evaluated comparisons with other methods, namely K-means, fuzzy K-means, principal component analysis, regularized expectation maximization and fuzzy C-means algorithms, it has been proven that the proposed method performs well.