Parallel MRI has brought forward new applications by reducing the scan time in MR image acquisition. The acceleration is achieved by reducing the number of phase encode steps acquired during the image acquisition thus giving aliased images. Sensitivity encoding (SENSE) is a widely used method to unwrap the aliased images. One major problem is the noise amplification during the process of reconstruction as well as the artifacts, which may arise during the reconstruction process. Regularization is an important tool to reduce this noise amplification (measured by geometry factor [g-factor]) and the artifacts, but it is computationally expensive. This work aims at exploring an alternative to regularization that would not be computationally expensive. The alternative suggested uses wavelet denoising of the coil sensitivity maps before applying standard SENSE reconstruction algorithm. The results of this reconstruction are compared with the reconstruction obtained by Tikhonov regularization. The results show that for lower acceleration factors (AFs; 2 and 3), the wavelet denoising of sensitivity maps before applying standard SENSE produces comparable results to those of Tikhonov regularization at much lesser computational cost but for higher AFs Tikhonov regularization is a better choice to acquire good quality images.