Aiming at the problems that the fault current is weak caused by the neutral point grounded through arc suppression coil, and the fault characteristics are not obvious, which leads to slow and difficult line selection when a single-phase ground fault occurs in resonant earthed system, an image-based single-phase grounding fault line selection method in distribution network is proposed. This method obtains the zero-sequence current of each feeder fault, through continuous wavelet transform (CWT), the zero-sequence current obtains the CWT coefficient matrix, and then the coefficient matrix is generated into RGB images. Finally, the Hamming distance between RGB images generated by the normal feeder and the faulty feeder is calculated by perceptual hash algorithm, which is used to judge the difference between RGB images, and then the faulty feeder is determined. The line selection methods based on Pearson correlation coefficient and based on EMD decomposition of high frequency components of zero-sequence current are compared. The results show that the proposed method has fast line selection speed and is not affected by factors such as fault resistance, fault location and initial phase angle of the fault when applied to the resonant earthed system.