Nondestructive identification for gender of chicken eggs based on GA-BPNN with double hidden layers

Abstract

SUMMARY: In order to identify the gender of chicken eggs at the early stage of incubation, a machine vision image acquisition system was constructed. Under the light source of LED, the images of 2 batches (186 and 180) of chicken eggs were respectively obtained on d 3, d 4, d 5, d 6, d 8, and d 10 of incubation. Considering the clarity and the integrity of blood vessels in the field of machine vision, the image of d 4 was determined as the basis for gender identification of chick embryos. After image processing, the 11 dimensions of feature parameters depicting the chick's embryonic development were extracted. In this paper, the genetic algorithm (GA) was used to optimize the initial weights and thresholds of backpropagation neural networks (BPNN) with different hidden layers. Then the GA-BPNN with single hidden layer, as well as, double hidden layers was established respectively. According to the research, the comprehensive accuracy of GA-BPNN model with double hidden layers reached 89.74% for the prediction set, which was higher than that of the model with single hidden layer, indicating that optimizing the initial weights and thresholds of BPNN by GA and adding the hidden layer had a certain effect on improving the recognition accuracy. Meanwhile, the results showed that the machine vision technology provided a feasible method for gender identification of chicken eggs at the early stage of incubation.

Keywords

• machine vision
• genetic algorithm
• BP neural network
• early stage of incubation
• chicken eggs
• gender identification