An image photographing device including a lens holder including a lens group and a coil unit that generates an electromagnetic force, a housing into which the lens holder is movably inserted, including a magnet that faces the coil unit, an image sensor, and a diaphragm formed of nonmetal and attached to the housing and elastically supporting the lens holder. Accordingly, the diaphragm has a predetermined elasticity although not reducing the thickness, the lens holder can properly follow the signal of the coil unit, the image photographing device is highly producible, and the possibility that the diaphragm is plastically deformed is small. Also, the terminal of the coil unit is connected to a separate terminal unit installed outside the diaphragm, such that the material used to form the diaphragm can be selected regardless of the conductivity, and foreign materials, such as, flux, can be prevented from contaminating the inside of the device and degrading the quality of an image.