CAJ | 학술논문

针对经典随机游走分割的不足，本文提出了基于L0平滑的随机游走分割方法。利用基于L0梯度最小化的图像平滑方法对红外图像进行预处理，在此基础上对图像进行随机游走分割。实验结果表明基于L0梯度最小化的图像平滑方法能够降低局部图像灰度值变化对经典随机游走分割路径的影响，在很大程度上提高了经典随机游走对变电设备红外图像分割的准确性。
침대경전수궤유주분할적불족，본문제출료기우L0평활적수궤유주분할방법。이용기우L0제도최소화적도상평활방법대홍외도상진행예처리，재차기출상대도상진행수궤유주분할。실험결과표명기우L0제도최소화적도상평활방법능구강저국부도상회도치변화대경전수궤유주분할로경적영향，재흔대정도상제고료경전수궤유주대변전설비홍외도상분할적준학성。
Aiming at the problem of classical random walker segmentation,a segmentation method based on L0 image smoothing and random walker is proposed. The substation equipment infrared image is preprocessed by using the method of smoothing based on L0 gradient minimization and then segmented by random walker. Experimental results show that the segmentation method based on L0 image smoothing and random walker can reduce the influence of local image gray value change on the segmentation path of random walker,and can improve the accuracy of the classical random walker greatly on the substation equipment infrared image segmentation.