面向复杂地形的配电线路监控视觉图像检测  

Visual image detection of distribution line monitoring for complex terrain

作　　者：王飞翔 江翔 张玲玲 戴华冠 陈力 WANG Feixiang;JIANG Xiang;ZHANG Lingling;DAI Huaguan;CHEN Li(Jiangsu Electric Power Information Technology Co.,Ltd.,Nanjing 210009,China)

机构地区：[1]江苏电力信息技术有限公司,江苏南京210009

出　　处：《电子设计工程》2025年第7期76-80,共5页Electronic Design Engineering

基　　金：国网江苏省电力有限公司科技项目(XM202331160142)。

摘　　要：在复杂地形下,配电线路监控视觉图像检测易受到地域影响,导致图像检测结果与实际情况不符。为此,提出了面向复杂地形的配电线路监控视觉图像检测方法。构建视觉注意模型,采用KNN算法聚类图像样本,通过所有类相加判别准则,标注监控视觉图像检测区。使用低通滤波平滑处理图像,对图像进行过零检测,利用经典感受野和非经典感受野分析视觉激励的不同响应,计算视觉响应差异,获取最终突触电流,以此作为神经元输入,结合高级视皮层显著性结果,获取图像检测结果。由实验结果可知,该方法图像边缘轮廓清晰,且受到遮挡部分与实际一致。Visual image detection of power distribution line monitoring in complex terrain is easily affected by geographical factors,resulting in image detection results that do not match the actual situation.To this end,a visual image detection method for distribution line monitoring targeting complex terrains is proposed.Construct a visual attention model,cluster image samples using KNN algorithm,and label the monitoring visual image detection area based on the criterion of adding all classes.Use low-pass filtering to smooth the image,perform zero crossing detection on the image.Analyze the different responses of visual stimuli using classical and non classical receptive fields,calculate the differences in visual responses,and obtain the final synaptic current as neuron input.Combined with the high-level visual cortex saliency results,obtain the image detection results.According to the experimental results,the image edge contour of this method is clear,and the occluded parts are consistent with the actual situation.

关 键 词：复杂地形 配电线路 图像检测 突触电流 

分 类 号：TN25[电子电信—物理电子学]