Imaging Technique Hatches Egg-Sorting System

Brent D. Johnson

Chickens headed for market fall into three subclasses: broiler birds, egg layers and breeders. Broiler birds, or meat chickens, are drawn entirely from the male population, because they grow faster than females.

When a new group hatches, a percentage of males are selected to be reared separately, ensuring uniform body weight and simpler processing. Segregating baby chicks according to sex, however, requires workers to physically check every bird. This is labor-intensive and requires a certain amount of skill.

No one, least of all commercial chicken farmers, should count chickens before they hatch. But they can at least determine the gender of pre-emergent chicks with the help of two Ethernet-capable cameras.

To streamline this inspection stage, Embrex Inc. developed an automated gender-sort process to determine sex before the eggs are hatched. The company adapted its Inovoject platform, which automatically inoculates each egg against various diseases, to also draw a small sample of the allantoic fluid from the egg and determine the sex.

When eggs are 16 to 17 days old, a sampling module turns them on their sides, draws out allantoic fluid and drops the samples onto a well plate. An assay module then dispenses into the wells a reagent that turns a vivid yellow when it reacts with estradiol, a hormone secreted by female chicken embryos.

Two Ethernet-capable Cognex In-Sight 1000C cameras serve as the system's vision sensor, said Alan Chalker, an applications engineer at Embrex. The company had evaluated imagers from another supplier, but encountered difficulties with interfacing discrete input and output. The challenge is that, with each snapshot, the system inspects 420 samples (210 per camera) every three seconds, which is too much bandwidth for serial communications.

"Most products don't have the range of interface options to develop the algorithms," Chalker said.

Embrex also found that, instead of red-green-blue, it was able to generate more complicated algorithms for hue-saturation-intensity color planes that allowed the applications developers to distinguish chromaticity from brightness. Pixel counting, for example, didn't work because of debris from the egg and inconsistent color of the sample wells. This was a key attribute of the In-Sight instrument. The ability to use both color spaces interchangeably made the imager useful under a variety of lighting conditions, regardless of the hue or saturation.