基于遗传神经网络的芯片图像分割

芯片图像分割是上芯机机器视觉系统进行图像处理的重要环节,分割效果直接影响下一步芯片信息的提取。采用遗传算法对BP神经网络的权值和阈值进行优化,避免网络陷入局部最优,使其在全局范围内获得最优解。通过选择实数编码和遗传算子算法控制参数,利用优化后的神经网络对芯片图像进行分割;并用模板匹配法得到分割图像的匹配度,计算其平均值和方差。最后,将该算法与传统分割算法的进行比较,该算法的分割效果优于传统分割效果。

Image Segmentation Based on Genetic Neural Networks

The segmentation of chip images is the important link in the image processing of the machine vision system.The segmentation effect will impact the extraction of the chip information.Genetic algorithm was used to optimize weights and thresholds in the BP neural network.So the net could obtain the global optimal solution and avoid being trapped in the local optimum.Through the selection of real-number-coding and genetic operator algorithm control parameters,the optimized net was used to segment the chip image...