Résumé : Phase aberration is one of the main contributors to image degradation in ultrasound imaging. Normalized-Cross-Correlation (NCC) is one of the most extensively studied techniques to estimate the arrival delay error and the aberration profile. \textit{However, the performance of NCC can be compromised when the data have a poor signal-to-noise-ratio.} Here we propose an iterative filtered NCC (f-NCC) method to estimate phase aberration in speckle regions from the noisy radio-frequency data of a single frame of synthetic transmit aperture (STA) image. First, two-dimensional filters were applied in the aperture and temporal domains to suppress noise and off-axis signals in the transmit beamformed data from an arbitrary speckle region. Second, an iterative method that included the phase aberration in both the transmission and receiving process was used to estimate the phase aberration. Third, the estimation were improved by correcting the edge effect with an iterative zero-padding technique. Last, in the experimental study, we proposed and validated a formula to estimate the arrival time error caused by a sound speed error. The results from both the numerical simulations and experimental studies confirm that the proposed filtered NCC method can estimate the phase aberration profile correctly from noisy STA RF data in a speckle region. Image quality in terms of contrast, resolution and image signal-to-noise-ratio was improved. Potential applications of the proposed method to map local speed of samples are also discussed.