Identification of eggshell crack for hen egg and duck egg using correlation analysis based on acoustic resonance method

This work presents a novel method for eggshell crack timely detection of hen and duck eggs based on the acoustic resonance analysis. The developed experimental device consisting of an excitation device, an acquisition module of acoustic signals, and a personal computer was designed to generate and acquire the response signals by impaction on eggshell. The parameter of Pearson's correlation coefficients (PCCs) was calculated to compare the similarity between intact and crack eggs signals with various sampling points. The 200 sampling points of the response signal were selected as an effective response signal, which observed the values of higher than 0.85 and lower than 0.8 for intact eggs and crack eggs, respectively. Thereafter, multivariate analysis of variance (MANOVA) was conducted to find that the egg individual and the impact location were two remarkable influencing factors to the response signal from intact egg. The cross-comparison of signals from the same direction of each egg sample was proposed to get the PCCs and served as feature parameters. A universal linear discriminant function was built to distinguish intact and crack egg. In the conducted experiments, a crack detection level of 95.5% and a false rejection level of 5% were achieved using the mixed samples of 100 hen eggs and 100 duck eggs. From the findings, it can be concluded that the proposed method will assist in simplifying the classification algorithm and in advancing the applicability for multiple categories of egg for an online detection system. Practical Applications Eggshell crack produced during packing and transportation may lead to significant economic loss and food security problem. This article presents a novel method of eggshell crack detection based on the acoustic resonance analysis for different categories egg. The effective response signals selection method and remarkable influencing factors analysis method of response signal was conducted to weaken the random interference of noise and reduce the computation consumption. As a result, a universal linear discriminant function was built to distinguish the intact and cracked eggs, which can satisfy the requirement in sorting industry. The overall results sufficiently indicated that the proposed methods in this study have significant potential for online detection of eggshell crack with high throughput.