考虑光照条件的蛋壳蛋心颜色识别方法

An Improved Method of Eggshell Color Recognition

研究了机器视觉识别中光照条件对蛋壳蛋心颜色的影响;根据白壳与青壳蛋透光性的差异,对比分析了鸭蛋不同壳色在不同光照强度下的颜色信息,找出了白壳蛋与青壳蛋图像信息差异最大的参数以及对应的光照条件。同时,分析了与蛋黄颜色相关性较大的特征参数及光照条件。实验表明:光照条件的亮度等级为120时,鸭蛋图像的可分度最好,利用形心区域的R分量来判断蛋壳的颜色,计算量减少,分类准确,易实现;白壳蛋在180亮度及青壳蛋在210亮度等级下,根据形心区域的H分量建立判别模型,预测精度高。

The light conditions on the impact of egg shell color and egg yolk in machine vision recognition are studied. The penetrate percentage of the white eggshell and the cyan are different on the same irradiation condition, which would led synchronous difference in the image. The best irradiation condition was found on which the penetrate percentage difference was most distinct. Besides, the most relativity between yolk color and image information was also analyzed. Experiments show that： when brightness of light conditions is 120, the divisibility of the duck image is best. Mathematical model, which is based on R value of duck image centroid to determine the egg shell color, reduces the calculation, recognizes accurately and is easy to achieve. When white egg shells is in 180 brightness and cyan shell eggs is in 210, mathematical model, which is based on H value of duck image centroid region, is established. Its forecast precision is high.