Abstract:Aiming at the problem of noise interference in egg crack detection process,this paper collects the audio vibration signals of the eggs on the transportation. Drawing recurrence plot(RP) of audio vibration signals which are unprocessed and using recurrence quantification analysis(RQA) to extract the quantitative feature parameters of recurrence plot. These quantitative feature parameters are recurrence ratio,determinism,laminarity,entropy and maximum diagonal length. Using these parameters to detect whether eggs are cracked. Results showed that the accuracy of detection and classification of egg with cracks is verywell via a support vector machine (SVM),back propagation neural network(BPNN) models. 300 eggs were detected in this study. The results showed that the SVM model was better, in the SVM model,the recognition rate of intact eggs and crack eggs was 93.98% and 95.52%.