classification of parus strawberry fruit by combining image processing techniques and intelligent methods

Nowadays, in modern agriculture, the combination of image processing techniques and intelligent methods has been used to replace humans by smart machine. there is no international or domestic standard classification for parus variety. in this study, an artificial image processing and artificial neural network (ann) method was used to classify strawberry fruit of parus variety. in the first step, the fruit was divided into 6 classes (ann outputs) by the expert, and 100 samples were randomly collected from each class. in the next step, the images of the samples were captured and three geometric properties with twelve color properties (as ann inputs) were extracted. morphological and color characteristics of the area of the fruit were extracted using the functions defined in matlab software. optimum artificial neural network structures (15186) considering root mean squared error (rmse) and correlation coefficient (r2) were investigated to classifiy the strawberry samples. finally, the perceptron neural network with a structure of 61815 was selected with the classification rate of 83.83%. the results obtained by ann showed that the lowest and highest accuracy was related to small ripe (about 65%) and small raw (about 100%). based on the accuracy of the results and economic aspects, ann method is considered a proper method to classify strawberry fruit.

Classification of Parus strawberry fruit by combining image processing techniques and intelligent methods

Nowadays, in modern agriculture, the combination of image processing techniques and intelligent methods has been used to replace humans by smart machine. There is no international or domestic standard classification for Parus variety. In this study, an artificial image processing and artificial neural network (ANN) method was used to classify strawberry fruit of Parus variety. In the first step, the fruit was divided into 6 classes (ANN outputs) by the expert, and 100 samples were randomly collected from each class. In the next step, the images of the samples were captured and three geometric properties with twelve color properties (as ANN inputs) were extracted. Morphological and color characteristics of the area of the fruit were extracted using the functions defined in Matlab software. Optimum artificial neural network structures (15186) considering root mean squared error (RMSE) and correlation coefficient (R2) were investigated to classifiy the strawberry samples. Finally, the perceptron neural network with a structure of 61815 was selected with the classification rate of 83.83%. The results obtained by ANN showed that the lowest and highest accuracy was related to small ripe (about 65%) and small raw (about 100%). Based on the accuracy of the results and economic aspects, ANN method is considered a proper method to classify strawberry fruit.