摘要
本研究将计算机视觉技术与遗传神经网络相结合,建立一套适合于孵化鸡蛋可成活性自动检测的计算机视觉系统,通过计算机视觉技术获取了孵化鸡蛋的色度直方图,并提取了孵化鸡蛋表面颜色特征。采用遗传算法优化了多层前馈神经网络的拓扑结构与权值,提高了神经网络的学习质量和学习速度,实现了孵化鸡蛋可成活性的自动检测。实验结果表明,该方法准确率较高,并具有鲁棒性和高速度。
A genetic neural network is used in a computer vision system. In this system, detecting the fertility of hatching eggs is discussed.In the paper, the color histogram of egg's surface is obtained, and the related feature is extracted by the system. Structures and weights of multi-layer feedback forward neural network are optimiged using genetic algorithm,learning quality and speed of the neural network are improved, and automatic detecting the fertility of hatching eggs is achieved. The results of experiment show that this algorithm is robust,and possesses highly accuracy and fast processing speed.
出处
《计算机应用与软件》
CSCD
北大核心
2001年第6期5-10,共6页
Computer Applications and Software
基金
吉林省科学技术委员会科学技术研究基金(编号:980535)
