To address the shortcomings of traditional damage test and detection methods in terms of accuracy and practicability， the fracture characteristics and dimension extractions of cracks on key parts of bridge based on fractal theory and image processing technology is proposed. By establishing the functional relationship between the fractal box dimension of cracks and typical mechanical parameters， the rapid and accurate damage assessment can be achieved. In order to verify the effectiveness of this method， the fatigue loading test of T-type reinforced concrete simple supported beam was carried out. The fractal theory is used to establish the function between the fracture fractal dimension and the measured parameters such as deflection of mid-span， first-order frequency， static stiffness and dynamic stiffness under different fatigue loadings and damage states. And then， the matching relationship between fractal dimension and damage index was clarified. Results show that the damage assessment based on fractal dimension has a good accuracy and robustness， and the refined damage assessment can be realized according to the traditional physical parameters and the fractal dimension of cracks.