The principle and effect of image stabilizer for improving the definition of image are studied. First, the influence of shaking on imaging is analyzed according to the light path of camera and the expression of offset d caused by shaking is given; second, the paper establishes the control model of the image stabilizer, uses the proportional-integral-derivative (PID) control algorithm and gets the result of anti-shaking effects with the simulation of MATLAB; last, some objects are shot by a digital camera respectively under the conditions of opening and closing image stabilizer. Two groups, 8 images of 150112 pixels are obtained. Based on the histogram in frequency domain, the image shot with image stabilizer has 10 more high-frequency components than the image shot without image stabilizer on average; based on the energy gradient curve in spatial domain, the order of magnitude of the image shot with image stabilizer increases from 104 to 105.