In response to the problems of large training scale and high computational cost of unmanned aerial ve-hicle(UAV)aerial image segmentation algorithms in complex backgrounds,this paper proposes a new method for image segmentation using a coupled neural P system(CCNP).This method is based on the pulse mechanism and coupling mechanism of the coupled neural P system,integrating region growth with the coupled neural P system.Through the pulse mechanism of neurons,each pixel is automatically activated,and according to the activation status,pixels are incorporated into the same region,effectively improving the ignition speed of each pixel and o-vercoming the problem of difficult ignition due to low pixel values.Using Berkeley images for validation,CCNP performs better on IOU and DICE metrics with mean values of 0.94 and 0.97,and MAE metrics with mean values of 0.04.These three metrics can improve the accuracy of image segmentation and demonstrate the feasibility of CCNP.In addition,by analyzing the example results of Chinese power line insulator images and drone aerial ima-ges,the effectiveness of CCNP in segmenting insulators in aerial images under complex backgrounds is further demonstrated,providing a new method for segmenting aerial insulator images.
关键词
绝缘子/图像分割/耦合神经P系统/区域生长
Key words
insulator/image segmentation/coupled neural P system/region growing
维普
万方数据