Speckle noise is inevitably involved in fringe pattern in electronic speckle pattern interferometry (ESPI), removing the noise is an important issue in ESPI. By exploiting the directionality of speckle pattern, an improved spin filtering is presented: four fuzzy directional windows in the field of current point are defined, so as to turn the determination of accurate one-dimensional directional window is turned into the determination of fuzzy directional window, and then the adaptive weighted mean filtering is applied instead of traditional median filtering is applied as denoising in the determined direction. The performance of the improved spin filtering is analyzed by using both computer-simulated fringes and experimental fringe. Bilateral filtering, wavelet subband removal method are introduced for making a comparison and validation in the image processing effect. The experimental results show that the proposed method is effective in noise removal and edge preservation for ESPI images.