Interferometric Synthetic Aperture Radar (InSAR), as a modern geodetic technique, plays an important role in earthquake source studies, epicenter determination, and loss evaluation. In order to meet earthquake emergency response demands, it is necessary to develop fast and efficient InSAR data processing programs. In the InSAR data processing chain, image registration is one of the most time‐consuming operations. In this article, we discuss our application of parallel computing technology to optimize the image registration program xcorr for the GMTSAR package. We developed two parallel programs; these are the xcorr2 program, which runs on multicore central processing units (CPUs), and the xcorr2_cl program, which runs on graphics processing units (GPUs). Results show that both parallel programs significantly improve the computational efficiency of image registration and ensure the subpixel registration accuracy. Meanwhile, our parallel programs can be integrated into the GMTSAR package to accelerate InSAR data processing on individual desktop workstations with multicore CPU and GPU.