Automatic Classification of South Indian Regional Fruits using Image Processing

摘要

Objectives: The main objective of proposed system is to classify different kinds of South Indian regional fruits. The fruits classified based on Extraction of morphological and Fourier features of a fruit image by applying DTNB classifier. Methods/Statistical Analysis: The proposed method is adapted to achieve fruit classification. The digital image of any fruits was given as an input to the system. Background elimination is the first step employed, is given based on the threshold technique. This helps in extracting only the interested pixel regions. Noise of the cropped image was removed; by applying mean filter.Statistical, morphological and Fourier features extracted from the image for the classification Findings: According to the literature survey conducted, most of the researchers used SVM (Support Vector Machine), neural network, KNN classifiers etc. for Automatic classification of fruits. Most of the authors extracted either spatial features or Fourier features for the classification. Very few researchers extracted both spatial and Fourier features to classify the fruit. The proposed method extracts both spatial and Fourier features of the fruits, which are commonly available in south Indian regions. The proposed system uses a hybrid combination of Decision table and Na?ve Bayes classifier to obtain the accuracy of 88.08%. Mat Lab is used for extracting the features of fruit images. There is no sufficient work done on fruit image classification for south Indian image fruits. The results of the proposed work are above the average and found to be satisfactory in classifying the fruits. Application/Improvements: This proposed system further enhanced to recognize the sub categorization of a specific fruits. For example, mango further classified into Banganpalli, Alphonso, etc.