基于改进投影寻踪技术和模糊神经网络的未受精种蛋检测模型

Detection model of un-fertilized egg based on improved projection pursuit and fuzzy neural network

考虑基于计算机视觉技术的未受精种蛋无损检测中特征参数之间关系的复杂性和模糊性等因素,提出了一种基于改进投影寻踪模型和模糊神经网络相结合,作为种蛋是否受精品质归属的决策系统.利用量子投影寻踪技术对种蛋图像的复杂形状特征向量进行提取降维,给出了计算最佳投影方向的一种改进量子遗传算法;并且利用模糊神经网络的自动学习决策推理规则,实现了种蛋是否受精品质归属的无损检测.结果表明:该模型速度快且稳定,精度高且鲁棒性好,简单易于实现,精度达到99.37%,满足实际检测要求.

To overcome the complexity and ambiguity of non-destructive detection characteristics for unfertilized eggs based on computer vision technology, an improved projection pursuit method combined with fuzzy neural network was proposed to decide whether the eggs were fertilized. The feature vector of complex shape of egg images was extracted to reduce the dimensionality by quantum projection pursuit technology. An improved quantum genetic algorithm was established to calculate the best projection direction. The non-destructive detection method was realized to diagnose whether the eggs were fertilized based on automatic decision inference rule of fuzzy neural network. The results show that the proposed method can meet actual testing requirements with high speed, good stability and robustness, and the accuracy can reach 99.37%.