In the method of identifying phase distortion based on energy distribution,the quality of fringe image is one of the main factors affecting the accuracy of distortion recognition.In order to accurately distinguish the phase distortion,the fringe image needs to be filtered to filter out noise and background information in the image.An adaptive image filtering method based on energy distribution is proposed,which converts an image into an energy distribution map by Fourier transform.The threshold of wavelet filtering is adjusted according to the difference between the dispersive and undistorted image energy dispersion.It is intended to seek a threshold with the largest difference in energy dispersion as an acceptable threshold for the filtering method for the current image.Experimental verification is performed on single spatial domain,single frequency domain and dual domain filtering methods.The results show that the energy distribution between the distorted image and the undistorted image is between 1.4 % and 3.2 % in a single spatial or frequency domain filtering process.The non-adaptive dual-domain filtering method is used for filtering,the energy distribution is more than 7 %,while the adaptive dual-domain filtering method is used for filtering,and the energy distribution difference exceeds 8 %.The adaptive dual-domain filtering method significantly improves the energy recognition accuracy.