A Method of Grading of Mozafati Date, Using Image Processing

Iran has not yet gained the desired place in world date market, due to not considering the world standards in the processes of grading and packaging. Machine vision and image processing are two of the new techniques that have found their places of application in agricultural industry. In this research, a solution has been introduced for classification of Mozafati date introducing 5 different grades namely: intact, shrinkage, damaged, mouldy and splotchy. A number of images were initially taken from the dates. They had been chosen, divided, and graded by an experienced person in this field. Through image processing technique and artificial neural network in MATLAB, a method for recognition of dates' defective areas was prepared. By preparation of an algorithm, date grading was carried out by using two images from each date. Machine vision comparison was employed using standard criteria. Correct recognition rate in image processing method  for intact, shrinked, damaged, mouldy and splotchy dates amounted to 95.83%, 88.89%, 64.28%, 80.55% and 80.00% respectively.  First order polynomial equations were employed to define the function between dependent variable (weight of date) and the independent variables (length, diameter and area) in the process of fruit grading. The submitted model benefits from a good correlation in estimating the weight of the intact dates (R²=0.93).