Fourier ptychographic microscopy (FPM) is a newly reported techniques that bypasses the SBP barrier of conventional microscope platforms, which gets high-resolution (HR) images with large FOV. FPM uses an LED matrix as the illuminating source of the microscope. Each lighted LED corresponds to a low-resolution (LR) image. An HR image is generated from a set of LR images by FPM. Larger illuminating angle provides higher frequency information for the HR image. Therefore, FPM increases the NA of the low-NA objective lens while maintaining the large FOV. However, the process of FPM is usually time-consuming, since typically hundreds of LR images are recorded and equally involved in the iteration to maintain the quality of reconstruction. In this paper, we proposed a method to accelerate FPM reconstructing process, called Adaptive-FPM. Inspired by the concept of “keyhole imaging” in MRI, we set an energy change threshold in the reconstruction for each LR image to decide whether the image can be skipped in current iteration or not. In this way, some images will be skipped in further iteration, and the total reconstruction time can be reduced. The method was tested by both simulated data and biomedical data, which showed that the new method led to similar results with the original FPM method, while the run-time was reduced a lot.