Yang Jiajie, Xuan Lin, Huang Guoxuan, Huang Anning, Liu Qi, Xu Guiyun, Zheng Jiangxia
National Engineering Laboratory for Animal Breeding and MOA Key Laboratory of Animal Genetics and Breeding, College of Animal Science and Technology, China Agricultural University, Beijing, 100193, China.
National Engineering Laboratory for Animal Breeding and MOA Key Laboratory of Animal Genetics and Breeding, College of Animal Science and Technology, China Agricultural University, Beijing, 100193, China.
Poult Sci. 2025 Jul 3;104(11):105530. doi: 10.1016/j.psj.2025.105530.
Eggshell moist spots are a common eggshell defect that reduces consumer purchase acceptance of eggs. Current detection methods typically use single-sided dark-field imaging to assess the ratio of sum of spot areas to sum of shell area (RSS), but overestimate moist spot severity compared to consumer perception under natural lighting, leading to commercial waste. Therefore, accurate detection methods aligning with consumer visual perception are crucial for optimizing commercial egg grading. This study aimed to verify the representativeness of single-side imaging and to develop and validate a bright-field automated identification method to ensure the accuracy of RSS. 510 pink-shell eggs were detected. First, the symmetry of moist spots on eggshell was assessed. Results showed the distribution of moist spots on both sides of the image bounded by the long axis of the egg was significantly symmetrical for both under dark-field (r = 0.979, P < 0.001) and bright-field (r = 0.952, P < 0.001), confirming single-side imaging representativeness. Second, a bright-field automated method was established using optimized threshold, background subtraction, and feature filters. Comparison of the RSS by bright-field images (RSSb) and dark-field images (RSSd) revealed a significant difference (P < 0.001). This indicates that dark-field imaging could not accurately reflect the true RSS under bright-field conditions. The limitations of RSSd were further analyzed using segmented linear regression. The results showed that when the severity of eggshell moist spots was high, RSSd was greater than 7.12 %, which was significantly correlated with the RSSb (r = 0.969, P < 0.001). However, when the severity of eggshell moist spots was low, RSSd was less than 7.12 %, the correlation became smaller (r = 0.498, P < 0.001), and RSSd could not evaluate the true RSS. This meant that the old method based on dark-field images could not accurately reflect the degree of eggshell moist spots and bright-field image method should be used in future.
蛋壳湿斑是一种常见的蛋壳缺陷,会降低消费者对鸡蛋的购买意愿。目前的检测方法通常使用单面暗场成像来评估斑点面积总和与蛋壳面积总和的比值(RSS),但与自然光照下消费者的感知相比,高估了湿斑的严重程度,导致商业浪费。因此,与消费者视觉感知相符的准确检测方法对于优化商业鸡蛋分级至关重要。本研究旨在验证单面成像的代表性,并开发和验证一种明场自动识别方法,以确保RSS的准确性。对510枚粉壳蛋进行了检测。首先,评估了蛋壳上湿斑的对称性。结果表明,在暗场(r = 0.979,P < 0.001)和明场(r = 0.952,P < 0.001)条件下,以蛋的长轴为边界的图像两侧湿斑的分布均具有显著对称性,证实了单面成像的代表性。其次,利用优化的阈值、背景减法和特征滤波器建立了一种明场自动方法。明场图像(RSSb)和暗场图像(RSSd)的RSS比较显示出显著差异(P < 0.001)。这表明暗场成像不能准确反映明场条件下的真实RSS。使用分段线性回归进一步分析了RSSd的局限性。结果表明,当蛋壳湿斑严重程度较高时,RSSd大于7.12%,与RSSb显著相关(r = 0.969,P < 0.001)。然而,当蛋壳湿斑严重程度较低时,RSSd小于7.12%,相关性变小(r = 0.498,P < 0.001),且RSSd无法评估真实的RSS。这意味着基于暗场图像的旧方法不能准确反映蛋壳湿斑的程度,未来应采用明场图像法。