Face recognition is an active research area because of the importance of its application areas. Face recognition as almost all image processing based systems begin with acquired images elementary processing. The elementary processing is to enhance and/or highlight features. The elementary processing phase sets up ceiling on the recognition ability since any lost feature cannot easily recovered in consequent steps. Also, false features add ambiguities to the problem and may lead to wrong conclusions. This study proposes basic filters oriented towards face feature extraction. The proposed filters are basic natural non- square, BANAN, directive-symmetric filters. So, at each typical pixel point the filters provide four readings (North, South, East, and West).The readings measure the divergence from the point of study between the interior and the exterior in the specified direction. The filters experimental results show competitive performance compared to the Sobel and the Prewitt filters.