In this paper, we describe a hand gesture recognition engine based on Computer Vision (CV), as a computing platform to support gesture interaction between humans and computers. Presenting a simple approach to recognizing gestures through image processing techniques and a single video camera, we address the problem of generic hand gestures recognition, especially of spelled Sign Language hand poses, introducing a preliminary study to its kinetic component. In our methodology, the system initially removes the background of captured images, eliminating irrelevant pixel information. The human hand is then detected, segmented and its contours localized. From these contours significant metrics are derived, allowing a search in a pre-defined hand poses? library, where each pose is previously converted into a set of metric values. We discuss several algorithmic options to support our methodology and present experimental results, regarding the recognition of Portuguese Sign Language signs. We discuss future directions of our work.