In order to ensure the efficient operation of wind turbine generator system (WTGS) and the safety and stability of wind farms, it is necessary to promptly detect and repair the blade damage. The traditional methods of detecting WTGS blade damage mainly rely on manual inspection, which is time-consuming, laborious, and has low accuracy. Therefore, it is of important practical significance to study the damage identification method of WTGS blades based on image processing technology. Due to the drawbacks of existing methods, this research aimed to study the damage identification method of WTGS blades based on image processing technology. A method of expanding blade damage samples based on the improved Deep Convolutional Generative Adversarial Networks (DCGAN) was first proposed, which generated a high quality damage image sample set to improve the classification performance of the deep learning model. For the problem of damage images often affected by noise and environmental factors in practical scenarios, it was solved by morphology-based blade damage edge enhancement. In addition, the blade damage state evaluation and classification process based on multifractal spectrum (MFS) was provided. Finally, the experimental results verified that the proposed algorithm was effective.