Given the high labor intensity, low efficiency and poor accuracy of the artificial lighting method in the detection of biological vaccine strain, embryo egg, activity, a bionic embryonic egg activity image nondestructive detection system was designed. Considering the blood vessel extraction was affected by factors such as the uneven gray level of the embryo egg image and the uneven eggshell quality of the embryo egg, an adaptive threshold method based on the minimum class inner exponential variance was proposed to segment the image and extract the blood. Arterial inclusion of egg embryo imaging was a critical factor for the accuracy of the fertility detection. By analyzing the development process and lighting experiment of egg embryo, it was found that the method shooting light at the bigger part and imaging from the side was an optimal way for the egg embryo imaging. In embryo egg image acquisition, the brightness of the image was varied under the influence of the camera environment and the image acquisition system. In view of this problem, the V component was used to realize automatic image classification, and then the image process was performed respectively. The main blood vessel of the active embryo egg was constructed accurately through the image edge detection and mathematical morphology process of the denoised embryo egg image, and the embryo egg activity was determined by calculating the two values area percentage and area perimeter ratio of the main blood vessel in the embryo egg. Experimental results showed that the system can recognize one embryo egg with 0.093s, and the detection accuracy of embryo egg activity was 100%. This system can meet the accuracy requirement of the production of inactivated vaccine and freeze-dried vaccine.