Detection of Chicken Egg Fertility and Early Embryo Development Using 1 Hyperspectral Imaging

6 In this study, a non-destructive, rapid and accurate method was developed based on 7 hyperspectral imaging technology to determine the fertility and early embryo development of 8 eggs. A total of 170 white-shell chicken eggs including 152 fertile eggs and 18 infertile eggs 9 were used in this study and all eggs were incubated in a commercial incubator for 4 days. A 10 transmission-based hypercube with the wavelength range of 900-1700 nm was obtained for 11 each egg on each day of incubation. After imaging on every day, randomly selected 12 developing embryos were stopped by injecting sodium azide (NaN3). The region of interest 13 (ROI) of each hypercube was segmented and the image texture information was extracted 14 from the ROI. The K-means clustering algorithm was used on the spectral transmission 15 characteristics to classify fertile eggs, infertile eggs and dead embryos. A perfect detection 16 (100% accuracy) of fertility prior to incubation (Day 0) was obtained. The classification 17 results for all eggs on each day of incubation were 65.29% at Day1, 61.18% at Day2, 72.94% 18 at Day3 and 84.12% at Day4. The low classification results at Day 1 and Day 2 indicated that 19 the embryo development is hard to be detected during the first 2 days of incubation. 20