Recognition and Classification of Fruit Diseases Based on the Decomposition of Color Wavelet and Higher-Order Statistical Texture Features

Abstract

Most of the circumstances, it is to be considered that the forerunners of economic losses and production in the agriculture industry of a country are diseases of fruit. To condense the probability of such losses, early detection and removal of fruit diseases can be cogitated. Manual identification of defected fruit in the fruit surface is carried out manually employing human inspection which is very time consuming and laborious. Computer-aided automatic fruit disease recognition and classification system can reduce the cognitive burden of the traditional system and improve its efficacy of false revealing of defected fruit. In this paper, we have proposed a framework for fruit disease identification using segmentation techniques and supervised learning based on data analyzed from the decomposition of color wavelet and higher-order statistical texture feature. Three types of common apple diseases are taken into considerations in this paper. We evaluated our method that can segment the image and shows its output as a result of classified defected fruit. As there are many popular techniques for image segmentation, we have used k-means clustering because of its unsupervised learning mechanism and it can be used to segment the region of interest from the background. Segmented images are extracted from the decomposition of color wavelet with higher-order statistical texture feature and combined which are used to train a multi-class linear support vector machine (MSVM). Gray-level run-length matrix (GLRLM) is used for higher-order statistical texture features of different directions (Ɵ = 0°, 45°, 90°, 135°). The experimental result demonstrates that the proposed approach is auspicious which triumphs the best performance of accuracy of 94.95% for the classification of defected fruit.KeywordsSegmentationClusteringWaveletRun-length matrixSupport vector machine