Interactive interfaces and applications are a flourishing research area. In this paper, we introduce a head gesture interface for a digital camera shooting self-portrait pictures. Natural head nodding and shaking gestures can be recognized in real-time, using optical-flow motion tracking. A double head nod triggers the camera shutter to take shots. Continuous nodding or shaking triggers a zooming interface to zoom the user's face in or out. To make the recognition robust, a safe zone analysis of the head region was conducted to quickly exclude any insignificant head motion, and thresholds of moving direction and length of head motion were selected in a preliminary set-up step. A finite state machine was used to recognize head gestures. Our results show that the proposed head gesture recognition method is a promising interface for a self-portrait camera.