<p>Carbon nanotubes occupy an important position in nanomaterials due to their unique physical and chemical properties. However, there are still some challenges to its identification and analysis, especially in the context of complex contexts. Given this, this study first takes the optimization of the image orientation of carbon nanotubes as the starting point, and selects the best parameters through the calculus calculation of the characterization parameters, so as to arrange the spatial positions of carbon nanotubes in an orderly manner. Secondly, on the basis of the Otsu’s thresholding method, the seagull optimization algorithm is introduced to improve, and finally a carbon nanotube contour extraction and recognition model is proposed. Results showed that the maximum orientation accuracy of the proposed orientation optimization method reached 83%, which was about 15% higher than that of other methods. In addition, the average segmentation accuracy of the contour extraction and recognition model reached 94%, the maximum between-class variance was 0.96, the maximum regional consistency was 0.934, the shortest single carbon nanotube image segmentation time was 11.5s, and the minimum difference between the upper and lower thresholds was 1. In summary, the proposed model not only improves the accuracy of carbon nanotube identification, but also provides new ideas for image processing of other nanomaterials.</p>