To achieve laser remote sensing classification of helicopter, propeller and turbojet aircraft, a texture feature extraction algorithm of aircraft target based on time-frequency image is studied. Three types of aircraft rotating parts echo signal are simulated according to the rotor micro-Doppler model, and the grayscale image is generated by the time-frequency distribution obtained by smoothed pseudo Wigner-Ville transform. OTSU method combined with grayscale stretching is used to perform threshold de-noising on the image, and the gray-level co-occurrence matrix (GLCM) feature and the Tamura feature are extracted. Feature optimization is carried out for the time-frequency difference, and finally the support vector machine (SVM) is used to classify the aircraft targets. Simulation data classification results show that the GLCM feature is sensitive to noise performance. When the time-frequency image is denoised by the proposed method and the signal-to-noise ratio (SNR) RSN=0 dB, the classification correct rate reaches to 96.4%. The Tamura feature has a higher classification accuracy under high SNR conditions, but decreases significantly when RSN<5 dB. Therefore, good classification performance can be obtained with the extraction of the texture feature of time-frequency image, and accurate classification of targets can be achieved by the improved GLCM feature under low SNR conditions.