Egg freshness detection based on the fractal dimension of sound signals

Eggs are a widely consumed material that can endanger people's health due to their high perishability. Therefore, detecting its freshness is very important for the food industry. Several nondestructive methods have been proposed to measure the internal quality of eggs, and in this research, an acoustic system was used to detect egg freshness. The samples were kept at room temperature for 16 days. A 45 cm long inclined surface with a 4 mm high step in the middle was used in the mentioned system. The sound resulting from the rolling of the samples and the impact caused by passing the step were recorded by a microphone. Then, the fractal dimension of sound signals was calculated using Higuchi and Katz methods. A destructive test was also done considering the Haugh unit and the eggs were divided into two groups AA and A+B in terms of freshness. The results of the three classification methods of artificial neural network (ANN), support vector machine (SVM) and linear diagnostic analysis (LDA) showed that the SVM method has better efficiency compared to the other two methods. The average classification accuracy of test data in the SVM method for Katz fractal dimension and window length of 25 milliseconds was obtained as 78.07%.