Abstract
In order to eliminate the infertile, weak and dead embryo eggs, which would pollute the normal embryo eggs, the activity detection of embryos need to be carried out before vaccination and harvest of influenza vaccine. Light method is widely used to detect the thickness and lines of blood vessels of chicken embryo to determine the embryo activity in the detection of vaccine strain activity. However, long time light detection is labor intensive, inefficient and easy to cause visual fatigue, which would lead to erroneous judgement of embryo eggs, and long time leaving the incubator will affects the development of chicken embryos. Given the disadvantages of high labor intensity, low efficiency and poor accuracy in the biological vaccine strains activity detection by artificial lighting, a nondestructive detection system of bionic embryo egg based on visual sense was designed. Meanwhile, a new adaptive threshold image processing method based on the minimum class exponential variance is proposed to solve the problem of extracting blood vessels from egg embryo which is affected by different gray level in egg embryo image and low quality of egg shell. The binary pattern of active egg embryo blood vessels were constructed accurately by the edge detection and mathematical morphology processing of the de-noised egg embryo images, the egg embryo activity is determined by calculating the binary area percent and area circumference of egg embryo blood vessels. Through the collection of a large number of active egg embryo images and the analysis of blood texture structure by graphic construction method, the model database of active egg embryo images are constructed to provide a criterion for the rapid determination of egg embryo activity. Results show that this system can detect an embryo egg in 0.039 s and the accuracy rate is 100%, which can meet the requirement of inactivated vaccine and freeze-dried vaccine production. Extraction of vessel texture from egg embryo.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have