Diameter and depth of dilated blood vessels are two key parameters of port wine stain (PWS) lesions. Recently, optical coherence tomography (OCT) has demonstrated considerable promise for clinical test and the key parameters extraction of PWS. However, intensity attenuation in depth and speckle noise in OCT images constitute two primary limiting factors with respect to resolving the morphologic information of the PWS lesions. In order to enhance the visual quality of the OCT images, we develop new image speckle reduction algorithms for the OCT signal. In the study, epidermis segmentation is based on a dynamic programming scheme; according to the epidermal boundary curve as a baseline, the details in the deep are enhanced by the attenuation compensation. The visual inspection of the enhanced images is then improved by a new variation model that combines the regularization term with the statistical characteristic constraints (Rayleigh distribution) of data corrupted by OCT speckle noise to eliminate the corresponding multiplicative noise. The result shows that the proposed algorithm provides significant improvements of the edge information about the obtained PWS OCT images of dilated blood vessels, which is helpful to divide and extract the key parameters.