A method for recognizing particle tracks in photographic streamer chamber pictures was developed. The pictures were digitized by a high-resolution CCD camera and processed by a computer. After filtering track recognition was carried out in two steps: a "local" detection of straight track pieces in small sections of the image, followed by a "global" recognition of the complete track. All tracks were assumed to originate in a single, known interaction point and to be of circular shape. The procedure is demonstrated by means of streamer chamber data for high-energy nucleus-nucleus collisions obtained in the CERN experiment NA35.