Seismic imaging with active seismic sources can be used to image subsurface structures such as faults, fractures, and layer interfaces with higher resolutions than tomographic methods. However, for characterizing large‐scale three‐dimensional structures, oftentimes it is too expensive and thus impractical to conduct active seismic survey. Seismic imaging using passive sources has been used to image structures through the “pseudoactive” source imaging by assuming source information is known. However, in practice there exist some uncertainties on source information including location, origin time, source time function, and radiation pattern, which can inevitably affect the final imaging result. In this study, we propose a new source‐independent passive seismic imaging method based on reverse‐time imaging with converted waves and a new imaging condition. This method uses the interference between P and S waves at conversion points for structure imaging by back‐propagating seismograms recorded by all receivers. Since no forward modeling is needed in this method, source information is not needed. A new grouping imaging condition is proposed to improve the image resolution, in which seismic receivers are divided into several groups and the separated P and S wavefields of different groups are multiplied and then stacked over time to get an image. We test the new method on both synthetic and real data sets based on a downhole microseismic system that is used for monitoring a shale gas hydraulic fracturing treatment. The results show that the proposed passive seismic imaging method does not depend on knowing source information and can achieve higher imaging resolution.