The invention discloses an anti-countermeasure attack camera source recognition method based on local smooth projection. The method comprises the following steps: 1) preprocessing a camera image; 2) constructing a camera source recognition feature extraction network; 3) generating a noise image block set; 4) defining a local smooth projection loss function; 5) constructing a camera source recognition front defense network; 6) an application recognition model. According to the method, local smooth projection is utilized to enable the feature extraction process of camera source recognition to effectively suppress countermeasure noise so as to extract features with countermeasure robustness, and therefore countermeasure attacks can be defended in camera source recognition. Meanwhile, a camerasource recognition front defense network is adopted, the feature extraction process and the defense process are separated, training is easy, and the method can be migrated to different camera sourcerecognition networks. According to the invention, the accuracy, robustness and mobility of the camera source recognition method based on the deep neural network are considered.