Edge detection algorithms have traditionally utilized the Gaussian Linear Filter (GLF) for image smoothing. Although GLF has very good properties in removing noise and unwanted artifacts from an image, it is also known to remove many valid edges. To cope with this problem, edge preserving smoothing filters have been proposed and they have recently attracted increased attention. In this paper, we quantitatively compare three prominent edge preserving smoothing filters; namely, Bilateral Filter (BLF), Anisotropic Diffusion (AD) and Weighted Least Squares (WLS) with each other and with GLF in terms of their effects on the final detected edges using the precision/recall framework of the famous Berkeley Segmentation Dataset (BSDS 300). We conclude that edge preserving smoothing filters indeed improve the performance of the edge detectors, and of the filters compared, WLS yields the best performance with AD also outperforming the GLF.