In this paper a feature vector, which has been used for curve coding, is evaluated in case of a signature verification scheme using a real time digital signal processor. The feature extraction method is based on morphologically processing the vertical projections of prescaled signature images. Coding of the curve profiles is carried out using morphological opening which explores the pixel allocation along the directions of projection. Various lengths for the structuring element are employed to increase feature discrimination capabilities. The method has been implemented using the Texas Instruments TMS320C6713 digital signal processor in conjunction with the MATLAB and SIMULINK software. The efficiency of the method is examined using a new signature database which comprises 70 writers. The classification approach uses the weighted distance as a similarity measure. An error rate lower that 0.02% is obtained for the case of random and zero effort forgery.