Non-destructive detection on the fertility of injected SPF eggs in vaccine manufacture

Abstract

A non-destructive detection system based on machine vision was designed for identifying the fertility of eggs prior to virus cultivation. Specific imaging system and detection algorithm were presented. A method based on smallest univalve segment assimilating nucleus was introduced to distinguish the high-brightness speckle noise pixels in egg images since the high transmittance of holes in eggshell. After speckle noise elimination and egg blood vessels restored, the multi-layer feature extraction method was used to acquire the binary image including main blood vessels. Finally, fertility judgment was evaluated by calculating the percentage that blood vessel area occupies in the ROI image. The experimental results show the method achieved identification accuracies of 97.78%, average identification time of 0.109s. It is proved that the proposed method is feasible for automatic and unattended fertility judgment of eggs in vaccine manufacture.