Light reflected by glasses etc. worn by a driver picked up in a facial image of the driver is reduced. With a vehicle-mounted camera (1) equipped with a first LED illuminating device (12) and a second LED illuminating device (13), each illuminating device is lit up alternately in synchronism with times of photographing facial images and facial images of the driver are taken at each lighting time. The vehicle-mounted camera then compares brightness for each pixel positioned at the same relative positions on each photographed facial image and extracts pixels of low brightness. The vehicle-mounted camera then generates new facial images for transmission to an ECU by synthesizing each of the extracted pixels. The ECU on the receiving side then detects inattentive driving etc. by the driver based on the facial images.
