In order to solve the online detection problem of micro-cracked eggs, this paper proposed a new nondestructive test method with multi-dimension features. This method combined acoustic feature and pressure feature to detect micro-cracked eggs. The external pressure was used to increase the micro crack. Firstly, the crushing experiments were carried out on 20 microcracked eggs with different cracks, and the characteristic curves of microcracked eggs under loading condition were determined. Associating with the mechanical property of intact eggs, the preliminary pressure range was selected. The selected pressure range was 0~6N. Furthermore, the audio signals of intact eggs and micro-cracked eggs were collected under the selected pressure range, vibrating by the sweep frequency band between 1~8000Hz. Through power spectrum analysis and principal component analysis, the optimal pressure was 5N for increasing micro cracks, and the optimal range of sweep frequency was 3000~7500Hz, respectively. And these parameters were suitable for the condition of industrial production line. In the experiment, 320 eggs were detected, and the detection model based on least squares support vector machine (LS-SVM) was constructed to detect microcracked eggs. This method was compared with the methods of BPNN and PNN. The results showed that the accuracy rates were 98.3% and 95% for intact eggs and micro-cracked eggs, respectively, and the detection time of each egg was 1s, the detection speed was 3600 eggs per hour. The proposed detection method is suitable for the online detection in assembly line.