雾霾场景下绝缘子缺陷检测模型的建立与仿真

Construction and Simulation of Insulator Defect Detection Model in Haze Scene

近年来,雾霾天气严重影响到绝缘子自爆缺陷检测的精度和电力系统的正常运行。以雾霾场景下自爆绝缘子图像为研究目标,采用暗通道先验去雾算法对图像作预处理,以避免绝缘子边缘检测结果断裂漏检现象的发生。针对绝缘子图像中的雾霾问题,运用直方图均衡化去雾算法、单尺度Retinex去雾算法、暗通道先验去雾算法进行仿真实验。结果表明,采用暗通道先验去雾算法得到的图像峰值信噪比平均值和结构相似性平均值均高于直方图均衡化去雾算法和单尺度Retinex去雾算法。

In recent years,haze weather has seriously affected the accuracy of insulator self-explosion defect detection and the normal operation of power system.Taking the image of self-exploding insulator in the haze scene as the research objective,the dark channel prior defogging algorithm is used for image pre-processing to avoid the occurrence of missing detection of insulator edge fracture.Aiming at the haze problem in insulator images,histogram equalization defogging algorithm,single scale Retinex defogging algorithm,and dark channel prior defogging algorithm are used to compare and analyze image defogging performance.The simulation results show that the average PSNR and SSIM of the image obtained by the dark channel prior defogging algorithm are higher than those of the other two mainstream defogging algorithms.