Oil–pressboard system is the main insulation form of oil-immersed transformers

which is prone to partial discharge (PD) and subsequent faults due to contamination or moisture ingress. To investigate the evolutionary characteristics of PD streamers under varying voltages and insulation dimensions

streamer images were captured using a high-speed camera system. An image restoration algorithm was proposed

which integrates the Retinex theory with bilateral filtering and k-dimensional tree (KD-tree)-based methods. This algorithm allows for the enhancement of image contrast while ensuring that edge details are preserved and fragmented streamer channels are reconnected. By varying the applied voltage

needle–plate gap length and pressboard thickness

patterns in streamer area and the number of bifurcations were analysed. These variations were microscopically explained in terms of electric field distortion and the dielectric matching between oil and pressboard insulation. It was found that larger streamer areas are produced by higher voltages and longer needle–plate gaps. Additionally

under elevated voltages or prolonged energisation

multiple streamer patterns are generated due to the presence of photoionisation and bubble formation. Furthermore

an increased number of bifurcations is observed with longer gaps and more uniform dielectric matching between oil and pressboard. The results obtained here enrich the evolution mechanism of PD and provide theoretical guidance for the dielectric matching of transformer oil and pressboard.